Design, development, and evaluation of an advanced cancer surveillance system

Abstract

Background

Cancer remains a leading global cause of mortality, demanding robust surveillance systems to inform public health strategies. Current cancer surveillance systems, particularly in low-resource settings, often lack on-demand analytics, spatial visualization, and predictive modeling, limiting their utility in addressing disparities and guiding targeted interventions. This study aimed to design, develop, and evaluate a GIS-integrated cancer surveillance systems tailored to the epidemiological and geographical context of Iran.

Methods

Employing a three-phase approach, the study began with a systematic review of cancer surveillance indicators, followed by the design and development of the system using a modular architecture supported by Django and Vue.js frameworks. The system integrates multi-level data standardization, GIS-based spatial analysis, and predictive analytics for on-demand insights. Usability evaluation was conducted using Nielsen’s Heuristic Assessment, incorporating feedback from medical informatics specialists, pathologists, and health managers.

Results

The Cancer Surveillance System incorporated critical data elements validated with CVR (> 0.51) and Cronbach’s alpha (0.849). Phase two developed a GIS-integrated, scalable system handling 20 million records, enabling on-demand monitoring, spatial analysis, and risk factor evaluation. Predictive modeling tools forecast cancer trends over 5-, 10-, and 20-year horizons, adhering to WHO standards. Usability evaluation resolved 85% of identified issues, enhancing functionality, user satisfaction, and scalability for precision cancer surveillance.

Conclusion

This study presents a scalable and adaptable CSS framework that bridges traditional surveillance limitations and modern analytical demands. Its integration of advanced technologies provides a model for global adaptation, supporting equitable resource distribution and evidence-based cancer control strategies.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12885-025-14947-7.

Introduction

Cancer remains one of the leading causes of morbidity and mortality worldwide, accounting for approximately 10 million deaths annually, as reported by GLOBOCAN [1]. The increasing global burden of cancer is driven by factors such as population growth, aging demographics, and lifestyle changes [2]. In Iran, cancer represents the third leading cause of mortality, following cardiovascular diseases and traffic accidents, posing a significant public health challenge [3]. Based on the Global Cancer Observatory, the incidence rate in Iran has increased from 107.3 per 100,000 individuals in 2008 to 141.6 per 100,000 in 2018, reflecting a sharp rise over the past decade [4]. Notably, gastric, lung, and breast cancers exhibit the highest mortality rates among Iranian men and women, respectively, with gastric cancer remaining the most prevalent in males [4].

Globally, cancer surveillance systems (CSS) are critical for public health decision-making, enabling the identification of trends, evaluation of interventions, and resource allocation [5, 6]. Despite advancements in healthcare technology and the implementation of CSS and cancer registries, substantial challenges persist in cancer monitoring, prevention, and control [5, 6]. However, limitations such as incomplete datasets, inadequate on-demand analytics, and poor geographic resolution hinder their efficacy [7–9].

The Iranian National Cancer Registry, established in 2014, marked a significant milestone in systematic cancer data collection. It integrates data from diverse sources, including pathology reports, hospital records, and vital statistics. Rigorous quality assurance measures are employed, such as structured reporting aligned with International Classification of Diseases for Oncology, 3rd Edition (ICD-O-3) standards, duplicate record elimination, and cross-validation across multiple data streams [7, 8]. In Iran, the current CSS, known as Simaye-Saratan, demonstrates limitations consistent with global challenges. It primarily depends on static reporting and fundamental descriptive statistics, lacking advanced analytical methods or spatial analysis capabilities. Additionally, the absence of Geographic Information System (GIS) integration restricts the ability to analyze spatial patterns of cancer incidence and identify high-risk regions, thereby limiting actionable insights for targeted interventions [7, 8].

To address these gaps, next-generation CSS must integrate advanced technologies such as GIS for spatial mapping, machine learning for predictive modeling, and dynamic dashboards for on-demand visualization [8, 9]. GIS has proven invaluable in public health surveillance, facilitating the identification of cancer hotspots, geographic disparities, and environmental risk factors [8]. While international systems like the Global Cancer Observatory (GCO) demonstrates the potential of these tools, their lack of regional adaptability and granular insights highlights the need for context-specific solutions [6, 10].

This study proposes the design, development, and evaluation of a GIS-based CSS tailored to the epidemiological and geographic context of Iran. By integrating on-demand analytics, spatial risk analysis, and predictive modeling, the system aims to overcome the limitations of existing platforms. It seeks to create a dynamic and adaptable platform capable of providing comprehensive insights into cancer indicators. This system aspires to bridge the gap between traditional cancer registries and modern surveillance platforms, empowering policymakers and public health officials with actionable data for cancer prevention, control, and resource optimization in Iran and potentially beyond.

Materials and methods

Study design

This study utilized a multi-phase, evidence-based methodology to design, develop, and evaluate a comprehensive CSS, structured into three interconnected phases: (1) requirement analysis and data collection, (2) design and development, and (3) evaluation. Figure 1 provides a concise, glanceable overview of the methodology, highlighting key processes in each phase. In phase 1, the research employed a descriptive methodology involving systematic literature and global CSS reviews and domain experts’ opinions via a researcher-developed requirement analysis checklist, followed by data analysis for identification of key system requirements. Phase 2 applied a practical development approach, leveraging the results of phase 1 to execute system design with Unified Modeling Language (UML) diagrams and system development through implementation of front-end, back-end, and database components. In Phase 3, a descriptive methodology was employed, utilizing a usability evaluation approach. Data were collected through usability testing conducted by a sample usability expert, guided by Nielsen’s heuristic evaluation checklist to systematically identify and analyze usability issues.

Fig. 1.

Categorization of the study’s methodology across three phases

Phase one: systematic review and requirements analysis

Systematic review literature

A systematic review was conducted in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guideline [11], to delineate critical data elements for a CSS, as documented in previous published study [12]. The search strategy yielded 1,085 articles from databases including PubMed, Embase, Scopus, Web of Science, and IEEE, covering the period from 2000 to 2023. Following a rigorous screening process, 13 studies were selected for inclusion, with a focus on epidemiological indicators, standardization practices, and system interoperability. The analysis revealed that key metrics such as incidence, prevalence, mortality, survival rates, Years Lived with Disability (YLD), and Years of Life Lost (YLL) are fundamental for effective cancer monitoring. Furthermore, the studies underscored the importance of adopting standardized classifications, such as ICD-O-3, and standard populations (e.g., SEGI, World Health Organization (WHO), or National standards for age-adjusted calculations), alongside demographic stratification by age, sex, and geography, to ensure enhanced data comparability and utility in cancer surveillance.

Evaluation of global cancer surveillance systems

To establish a foundation for the proposed CSS, a comparative evaluation of 13 international CSS was conducted to identify universal data elements, standardization practices, and best practices for data structuring, visualization, and reporting, ensuring alignment with international standards while addressing local epidemiological needs. These included GCO [10], European Cancer Information System (ECIS) [13], Cancer Research United Kingdom [14], Australian Cancer Data System [15], NORDCAN [6], United States Cancer Statistics Data Visualization Tool [16], National Children’s Cancer Registry Probe [17], Spanish Network of Cancer Registries [18], Dimensions of cancer [19], Finnish Cancer Registry [20], National Cancer Registry of Ireland [21], Geodes – French Public Health Agency [22], and the Hamid and Christina Moghadam Program in Iran Studies Health Dashboard [23]. The selected systems were chosen for their geographical diversity, accessibility, and comprehensive documentation, as detailed in our prior publication [12]. The evaluation focused on core epidemiological indicators, alongside standardization practices and visualization tools like heatmaps, time-series graphs, and GIS-based spatial analyses, assessing their applicability across diverse contexts. Results highlighted strengths in standardized metrics and advanced visualization features (e.g., user-friendly dashboards in GCO), but noted limitations in subnational granularity and data quality from low- and middle-income countries, which informed the design of the proposed CSS framework.

Development of a standardized data checklist

A standardized data checklist was developed, consolidating critical CSS elements identified in the systematic review and global evaluation. The checklist underwent rigorous validation, employing the Content Validity Ratio (CVR) to assess the necessity of each data element and Cronbach’s alpha to evaluate internal consistency [24, 25]. A diverse expert panel of oncologists, epidemiologists, and public health specialists from Zanjan University of Medical Sciences (ZUMS) participated in this process, ensuring statistically robust input and broad applicability.

Data collection

This study utilized both individual-level and aggregated data to develop the CSS framework. Individual-level data from the Iranian National Cancer Registry included patient-level records of cancer diagnoses, encompassing epidemiological metrics such as incidence, prevalence, mortality, survival rates, YLL, and YLD, classified using ICD-O-3 codes [26]. Data were aggregated from multiple sources to support risk factor analysis. These included annual statistical booklets, in PDF format, containing cancer and demographic statistics, extracted from the websites of authoritative statistical agencies, namely the Iranian National Statistics Center [27], and the Civil Registration Organization [28]. Additionally, air pollution data were retrieved from the Iran Air Pollution Monitoring System [29] through a Python-based query. Data preprocessing entailed the extraction of booklet data into Excel format, followed by deduplication to eliminate redundant entries, standardization to ensure consistency across data points, and classification into 57 distinct data items for subsequent analysis. Data were classified into five categories including cancer-related variables, socio-demographic variables, healthcare infrastructure variables, environmental conditions, and air quality (Table S1).

Phase two: system design and development

System design and architecture development

The CSS design process was methodically structured to address diverse user requirements, commencing with an end-user needs analysis to identify and prioritize specifications for system users. UML was utilized to develop data flow diagrams, ensuring robust data integration and interoperability within the CSS framework. Use-case diagrams delineated interactions between user groups and the system, specifying role-based access control (RBAC) and functionality. Sequence diagrams mapped workflows, detailing interactions among users, servers, and the database to facilitate intuitive and efficient task execution. Activity diagrams provided comprehensive representations of process flows, enhancing clarity in operational tasks. Class diagrams defined the database schema, organizing cancer patient data, epidemiological indicators, and analytical results into relational tables for efficient storage and retrieval (Figs. S1–S30). An Application Programming Interface (API) was implemented to enable seamless data exchange, supporting a responsive front-end for real-time interaction. A relational database was established to ensure efficient data management, enhancing the framework’s scalability and adaptability (Fig. S31).

Programming and development tools

The CSS was developed using Django (version 6.0.5) and Django REST Framework (version 13.2.3) for the backend, alongside Vue.js (version 5.4) for the frontend, leveraging Python (version 3.12.1), HTML, JavaScript, and MySQL (version 8.0) to establish a scalable, web-based client-server architecture. Django was selected over alternatives such as Flask due to the development team’s extensive expertise with its ecosystem, which provides a robust Object-Relational Mapping (ORM) system, integrated authentication mechanisms, and advanced security features. These capabilities facilitated the rapid development of a secure, database-driven health application without requiring proficiency in new technologies. For the frontend, Vue.js was chosen over React and Angular, capitalizing on the team’s familiarity, its straightforward syntax, and progressive integration features, which enabled the efficient creation of a responsive single-page application (SPA). The frontend incorporates dynamic visualizations through Chart.js and ApexCharts, with data export functionalities supported by xlsx and jsPDF libraries. A Representational State Transfer (REST) architecture ensures efficient JSON-based communication. The modular Django backend integrates libraries such as PySAL for spatial analysis, SciPy for statistical modeling, and pandas for data processing. Security is enhanced through RBAC, JSON Web Token (JWT) authentication, and Vue-Router guards. A comparative analysis of Django, Flask, Vue.js, and React, detailing their strengths and limitations, along with an evaluation of existing CSS frameworks, highlighting the advancements of the proposed system, is presented in Tables S2–S3.

Phase three: system evaluation

The CSS was evaluated using Nielsen’s Heuristic Usability Assessment Checklist, a validated framework designed to assess key usability dimensions, including ease of learning, efficiency, and error prevention [30]. This method, based on 13 predefined principles—such as system visibility, user control, error recovery, and flexibility, is widely recognized for its efficacy in identifying usability issues, even with a small sample of evaluators [31]. The checklist used in this study is provided in Supplementary Materials (S5). The evaluation was conducted at ZUMS with a panel of five experts, comprising medical informatics specialists, pathologists, and cancer registry managers, selected through convenience sampling. In accordance with Nielsen’s guideline of utilizing 3–5 evaluators to uncover the majority of usability issues, the process successfully identified over 85% of potential concerns [30]. Subsequent iterative refinements addressed these findings, enhancing the user interface and overall system functionality to ensure the platform effectively met its intended objectives.

Results

Phase one: development of a standardized data checklist

A validated checklist of essential data elements and filters was developed based on findings from the systematic review and global CSS evaluation. Core elements included incidence, prevalence, mortality, survival rates, YLL, and YLD, with standardized rates based on Iranian, SEGI, and WHO reference populations. Filters such as year, sex, age-group, geography, and cancer type supported granular analyses. Expert validation confirmed the checklist’s relevance and reliability. All elements achieved a CVR exceeding 0.51, ensuring their inclusion, and the checklist demonstrated strong internal consistency (Cronbach’s alpha = 0.849). Of the 17 distributed checklists, 14 were completed, with 85% male and 15% female participants. The participants represented a range of professional specializations, including pathologists (27.5%), oncologists (27.5%), health services management specialists (15%), epidemiologists (15%), and general practitioners (15%). Work experience varied, with 21% of participants having less than 10 years, 64% between 10 and 20 years, and 15% exceeding 20 years. The structured checklist, presented in Table 1, provides a validated core data elements for CSS.

Table 1.

Core data elements for cancer surveillance systems

Epidemiology Indicators	Essential N (%)	Necessary but Not Essential N (%)	Preferable N (%)	Not Necessary N (%)	CVR
Incidence	14 (100%)	0	0	0	1.00
Prevalence	14 (100%)	0	0	0	1.00
Mortality	14 (100%)	0	0	0	1.00
Survival	14 (100%)	0	0	0	1.00
YLL	8 (57%)	4 (29%)	2 (14%)	0	0.71
YLD	7 (50%)	4 (29%)	2 (14%)	1 (7%)	0.71
Filters
Crude Rates	6 (43%)	5 (36%)	3 (21%)	0	0.57
ASR (Iran)	8 (57%)	4 (29%)	2 (14%)	0	0.71
ASR (SEGI)	10 (72%)	3 (21%)	1 (7%)	0	0.85
ASR (WHO)	12 (85%)	2 (15%)	0	0	1.00
Time	14 (100%)	0	0	0	1.00
Gender	14 (100%)	0	0	0	1.00
Age Group	14 (100%)	0	0	0	1.00
Geographic	14 (100%)	0	0	0	1.00
Cancer Type	14 (100%)	0	0	0	1.00

Phase two: system design and development

End-User and system requirements analysis

The system was tailored for administrators, healthcare providers, and public health professionals. Functional requirements included data entry, advanced descriptive, statistical and spatial analyses, and flexible reporting. Non-functional requirements emphasized security, scalability, and usability, ensuring robust and reliable performance.

System development

The CSS was designed to address the diverse needs of its end-users by incorporating different modules, each tailored to fulfill specific functional and operational requirements. These modules reflect a comprehensive framework aimed at providing intuitive access, secure data management, and advanced analytical capabilities, aligning with the recommendations derived from global best practices and user-centric requirements.

Login and home page

The CSS employs a secure JWT-based authentication for authorized access, ensuring data protection. The Home Page, organized into five sections including Data Registration, Dashboard, Settings, Updates, and Help which streamlines navigation and aligns functionalities with user roles through RBAC, enhancing usability and security (Fig. S32).

Settings and user management

The Settings section facilitates secure management of user accounts, roles, and permissions, ensuring role-appropriate access (Fig. S33). A user search feature streamlines account management, while administrators can create accounts, assign roles, and define permissions. RBAC links permissions to roles, enhancing security and operational efficiency. These features safeguard sensitive data and ensure users can perform tasks aligned with their responsibilities, supporting the CSS’s objectives.

Data registration

The Data Registration section ensures accurate and systematic recording of patient information aligned with the standardized components of Iran’s National Cancer Registry (Fig. S34). Key data elements include demographic details (e.g., name, gender, national ID), geographic information (e.g., province, county), and clinical data (e.g., ICD-O codes, cancer grade, diagnostic method). Controlled input methods, such as dropdown menus, minimize errors and standardize data entry. The system integrates ICD-O-3 classification, filtering relevant morphological codes based on topological codes for precise cancer classification. Bulk data uploads in CSV or Excel formats streamline the integration of large datasets, reducing manual workloads while maintaining accuracy.

Dashboard

The Dashboard is the central analytical hub of the CSS, incorporating eight components to deliver comprehensive and user-friendly functionality. It includes geographical selection, a dynamic filtering system, a main menu for analysis types, a sub-menu for refining selections, visualization results in multiple chart formats, an export results section, the Anatomy view for organ-specific exploration, and a summary of selected filters (Fig. 2).

Fig. 2.

Interface and components of the cancer surveillance system dashboard

Geographical selection enables users to analyze cancer data at the provincial or county level, providing granular insights into regional distribution (Fig. 2-I). The dynamic filtering system refines analyses across dimensions such as epidemiological indicators (e.g., incidence, mortality, and survival rates), geographic regions, timeframes, sex, and age groups. It supports age-standardized comparisons using Iranian, SEGI, and WHO populations and leverages ICD-O-3 classifications for precise cancer type analysis, including anatomical systems, organs, and subtypes. Users can also forecast trends for 5, 10, or 20 years and explore multi-dimensional risk factor analyses (Fig. 2-II).

The primary interface menu of the CSS provides access to modules including Today’s Status, Trend Analysis, Forecasting, Statistical Analysis, Spatial Analysis, and Risk Factor Analysis (Fig. 2-III). Each main menu selection is accompanied by a corresponding sub-menu, offering detailed options tailored to facilitate customized analyses within each module (Fig. 2-IV). A range of visualization tools, encompassing bar graphs, line graphs, pie charts, and heat maps, is employed to effectively depict current status, temporal trends, and spatial disparities. Interactive maps at the province and county levels, developed in SVG format, along with dynamic charts, enhance different analysis capabilities (Fig. 2-V). The export functionality enables the sharing of data and visualizations in formats such as Excel, PDF, and images, thereby supporting reporting and academic research (Fig. 2-VI). The Anatomy View provides organ-based insights using ICD-O-3 classifications, visually highlighting affected organs through interactive color-coded schematics (Fig. 2-VII). A summary of selected filters ensures clarity, showing active filters applied to visualizations, enhancing transparency and user experience (Fig. 2-VIII).

Each component of the dashboard’s main menu is designed to deliver specialized insights and tools, facilitating comprehensive exploration of cancer-related data.

Today-Cancer Status offers on-demand updates through user-defined filters, summarizing key metrics such as incidence, prevalence, and mortality rates, and highlighting maximum values by year, cancer type, gender, age group, and location. Subsections like By-Location and By-Cancer use bar charts to reveal geographic disparities and rank the top 20 cancers using ICD-O-3 classifications, while the Age-Pyramid visualizes demographic patterns in 5-year intervals, identifying age-specific burdens. Pie and mosaic charts in the Cancer-Distribution-Percentage and Location-Distribution-Percentage subsections illustrate the contributions of cancer types and regions to the overall burden. An interactive Geographical-Map leverages advanced techniques like the Jenks Natural Break Classification algorithm [32] to highlight regions with significant trends, facilitating strategic interventions (Fig. S35).

Trend-Analysis complements this by examining temporal changes in cancer indicators, providing insights into patterns, intervention efficacy, and emerging trends across demographics and regions. It summarizes trends in total cases, incidence rate shifts, and demographic or geographic variations. Subsections like By-Location and By-Cancer use bar charts to identify regions with rising or declining indicators and prioritize top cancer types. The Age-Pyramid subsection visualizes age- and gender-specific trends, while the Geographic-Map highlights temporal changes spatially. Line-charts in the Trend-Analysis subsection detail patterns of increase, decrease, or stability in cancer metrics, offering a dynamic understanding of evolving indicators (Fig. S36).

Forecasting-Analysis predicts cancer indicators over 5-, 10-, and 20-year horizons, using historical data to calculate growth rates and forecast trends. These projections enable proactive planning and interventions (Fig. 3). The In-a-Glance subsection summarizes projected metrics, including case counts, incidence rates across standards, and the most affected demographics and regions (Fig. 3-A). By-Location uses bar charts to compare current and future rates by provinces or counties, highlighting regional trends (Fig. 3-B). By-Cancer forecasts trends for the top 20 cancer types, using color-coded visuals for comparison (Fig. 3-C). The Age-Pyramid subsection predicts cancer incidence by age group and gender across all horizons (Fig. 3-D). The Geographic-Map identifies future high- and low-risk areas (Fig. 3-E). The ISO-Chart provides an intuitive representation of projected incidence rates, with icons representing approximately 10 cases per 100,000, aiding resource alignment and strategic planning (Fig. 3-F).

Fig. 3.

Cancer surveillance system forecasting analysis pages interface

Statistical-Analysis is a core feature of the CSS dashboard, offering tools for examining cancer trends and supporting decision-making without external computations (Fig. 4). The In-a-Glance subsection presents key metrics, including average rates for various standards, range, variance, and interquartile range for trend stability and outlier assessment (Fig. 4-A). Advanced analysis includes the Interquartile-Range (IQR) subsection, which illustrates annual variability (Fig. 4-B), and the Relative-Risk (RR) subsection, comparing provincial rates to national averages with 95% confidence intervals (Fig. 4-C). The Temporal-Trend subsection analyzes variations in cancer incidence over selected time periods (Fig. 4-D). The By-Cancer-Percentage subsection assesses the top 20 cancer types with the highest prevalence percentages across all cancer types (Fig. 4-E). The By-Location-Percentage subsection delineates regions within the country exhibiting consistently high or low cancer incidence percentages (Fig. 4-F).

Fig. 4.

Cancer surveillance system statistical analysis pages interface

Spatial-Analysis is a vital CSS dashboard feature providing detailed geographic insights into cancer indicators through advanced tools, including spatial autocorrelation analysis (Moran’s Global Index), hotspot analysis, and cluster and outlier analysis (Moran’s Local Index). These tools identify statistically significant patterns, uncovering high-risk and low-risk areas across the country, with direct dashboard integration for accessibility and efficiency (Fig. 5). The Spatial-Autocorrelation-Analysis subsection visualizes provincial cancer incidence using scatterplots with each point representing a province. The vertical axis shows spatial lag (average incidence of neighboring provinces), while the horizontal axis reflects individual provincial rates. Clusters along a trend line indicate strong spatial relationships, whereas dispersed patterns suggest weak dependencies, aiding in the assessment of geographic cancer patterns (Fig. 5-A). The Hotspot-Analysis subsection identifies regions with significantly high or low incidence rates, employing techniques similar to those in GIS platforms like ArcGIS, categorizing them as hotspots or cold spots with confidence levels of 99%, 95%, or 90%. Utilizing Iranian provincial shapefiles, the PySAL library, and spatial statistical techniques, regions are classified into seven categories via color-coded confidence intervals. These insights guide targeted public health interventions (Fig. 5-B). The Cluster-and-Outlier-Analysis subsection (Moran’s Local Index) complements hotspot analysis by identifying localized clusters and outliers. Areas are categorized as HH (high-high clusters), LL (low-low clusters), HL (high outliers surrounded by low values), and LH (low outliers surrounded by high values). HH and LL clusters signify regions with homogeneous high or low cancer incidence rates, while HL areas highlight potential anomalies requiring further investigation. LH areas suggest protective factors or effective interventions (Fig. 5-C).

Fig. 5.

Cancer surveillance system spatial analysis pages interface

Risk-Factor Analysis examines the influence of various factors on cancer indicators, offering actionable insights for public health planning. Risk factors are grouped into population distribution, demographics, healthcare infrastructure, environmental conditions, and air quality. Analysis is presented through four subsections (Fig. 6). The Geographic-Map subsection uses color-coded intensity maps to depict spatial distributions of risk factors across Iran, identifying regions where specific factors may significantly influence cancer incidence. The Trend-Change subsection employs line graphs to analyze temporal variations in risk factors, revealing increases, decreases, or stability over time and their long-term impacts on cancer trends. The Correlation-Analysis subsection applies Pearson’s correlation coefficient, visualized on heatmaps, to explore relationships between risk factors and cancer indicators. Results, ranging from − 1 to + 1, are color-coded into eight levels, highlighting strong positive or negative influences (Fig. 6-A). The Regression-Analysis subsection uses multivariate Ordinary Least Squares (OLS) regression to evaluate the combined impact of risk factors. Scatterplots display regression coefficients and color-coded provinces, identifying significant risk factor influences across regions (Fig. 6-B).

Fig. 6.

Cancer surveillance system risk factor analysis pages interface

County-Level analysis

The system extends all provincial-level functionalities to the county level, enabling detailed spatial analysis (Fig. S37). Users can analyze data nationwide or focus on specific provinces using location filters. This granular approach identifies high-risk counties and trends obscured in broader analyses, supporting localized health strategies and targeted interventions for cancer prevention and control (Fig. 7). The By-Location-Percentage subsection identifies counties within the country or a selected province that exhibit consistently high or low cancer incidence percentages (Fig. 7-A). The Geographic-Map subsection visualizes high- and low-incidence rates by county (Fig. 3-B).

Fig. 7.

Cancer surveillance system county-level analysis pages interface

Help

A dedicated Help section, identified as a critical component through Nielsen’s heuristic evaluation, enhances user-friendliness and navigational efficiency for novice users. It features a hierarchical, interactive structure with concise descriptions and visual aids, promoting accessibility and enabling effective utilization of the CSS for cancer monitoring across users with diverse technical expertise (Fig. S39).

Phase three: usability evaluation

The usability evaluation of the CSS was conducted with five experts from ZUMS, encompassing a diverse group of professionals with extensive experience in cancer registration systems. The evaluators included 40% medical informatics specialists, 40% pathologists, and 20% health services management specialists, with 60% having less than 10 years of experience and 40% between 10 and 20 years. This diverse expertise provided a comprehensive assessment of the system’s usability and functionality.

The usability evaluation of the CSS framework identified 349 issues, with 57 deemed consistent across evaluators, leading to the consolidation of these into 121 unique concerns following a thorough review. The most frequent issues were categorized as follows: user interface respectfulness (64%, n = 77), user skill requirements (59%, n = 71), and error prevention mechanisms (60%, n = 73). Notably, no issues were reported regarding the privacy component, underscoring the robustness of the data protection measures (Table 2). Several enhancements were implemented to address usability issues and improve system functionality. The static PDF Help section was replaced with an interactive, hierarchical Help module, enabling users to efficiently locate information and understand system functionalities, thereby reducing frustration and enhancing the user experience. Descriptive error messages were introduced to clearly explain error causes and provide corrective suggestions. For example, users attempting incompatible queries, such as selecting male genital cancer for the female gender, now receive prompt guidance for adjustments. To address interface inconsistencies, a uniform color-coding scheme was adopted, improving coherence and reducing confusion. Functional clarity was further enhanced by resolving responsiveness issues, with responsive CSS coding ensuring proper scaling and separation of functional sections across different screen sizes. Advanced error prevention mechanisms were also integrated, including input validation before submission, automatic selection of standard values, default filter settings in dashboards, and dropdown menus to replace manual input. Collectively, these improvements significantly elevated the system’s usability, functionality, and user satisfaction.

Table 2.

Summary of usability issues identified through nielsen’s heuristic evaluation

Title	Evaluator 1 N (%)	Evaluator 2 N (%)	Evaluator 3 N (%)	Evaluator 4 N (%)	Evaluator 5 N (%)	Common N (%)	Non-Usable N (%)	Usable N (%)	Total N (%)
System Status Visibility	8 (27.6%)	5 (17%)	6 (20.7)	5 (17%)	6 (20.7)	4 (29%)	14 (42%)	19 (58%)	33 (11%)
System-Real World Alignment	7 (29%)	8 (33%)	6 (24%)	6 (24%)	5 (20.8%)	5 (42%)	12 (50%)	12 (50%)	24 (8%)
User Control and Freedom	10 (43.4%)	7 (30.4)	7 (30.4)	8 (34.8)	7 (30.4)	7 (64%)	11 (48%)	12 (52%)	23 (8%)
Consistency and Standards	12 (23.5%)	11 (21.5%)	14 (27.4%)	10 (19.6%)	11 (21.5%)	10 (55%)	18 (35%)	33 (65%)	51 (17%)
Error Recognition and Recovery	5 (23.8%)	7 (33.3%)	5 (23.8%)	5 (23.8%)	6 (28.6%)	5 (63%)	8 (38%)	14 (62%)	2 (7%)
Error Prevention	7 (46.6%)	5 (33.3%)	4 (26.6%)	5 (33.3%)	4 (26.6%)	4 (44%)	9 (60%)	6 (40%)	15 (5%)
Recognition over Recall	6 (15%)	4 (10%)	5 (12.5%)	4 (10%)	5 (12.5%)	4 (50%)	8 (17%)	32 (83%)	40 (14%)
Flexibility and Efficiency	4 (25%)	5 (31.2%)	5 (31.2%)	4 (25%)	4 (25%)	3 (42%)	7 (44%)	9 (56%)	16 (5%)
Minimalist and Aesthetic Design	2 (16.6%)	2 (16.6%)	3 (25%)	3 (25%)	2 (16.6%)	2 (50%)	4 (33%)	8 (77%)	12 (4%)
Use of Help	6 (26.1%)	5 (21.7%)	4 (17.4%)	4 (17.4%)	5 (21.7%)	4 (50%)	8 (35%)	15 (55%)	23 (8%)
Skills	8 (36.4%)	5 (22.7%)	5 (22.7%)	6 (27.3%)	5 (22.7%)	5 (38%)	13 (59%)	9 (41%)	22 (7%)
Interaction and Respect for Users	7 (50%)	4 (28.6%)	6 (42.8%)	4 (28.6%)	5 (41.6%)	4 (44%)	9 (64%)	5 (36%)	14 (5%)
Privacy	0	0	0	0	0	0	0	3 (100%)	3 (1%)
Total	82 (28%)	68 (23.2%)	70 (23.9%)	64 (21.8%)	65 (22.2%)	57 (47%)	121 (41%)	172 (59%)	293 (100%)

Discussion

This study introduces a significant advancement in cancer surveillance through the development and evaluation of a comprehensive CSS tailored to Iran’s specific needs. By integrating advanced technologies, robust analytics, and user-friendly interfaces, the CSS addresses critical gaps in existing systems, enabling on-demand monitoring, predictive modeling, and actionable insights to support cancer control strategies. The scalable, secure, and adaptable framework equips healthcare professionals, researchers, and policymakers with effective tools for data-driven decision-making. Drawing on lessons from pandemic-era dashboards the CSS incorporates real-time data aggregation, GIS-based mapping, and intuitive visualizations, adapted to the long-term management of non-communicable diseases, thereby enhancing the evolution of responsive public health infrastructure [33].

Phase one: systematic review and requirements analysis

The first phase underscores the need for a comprehensive cancer surveillance framework, integrating indicators like incidence, prevalence, mortality, survival rates, YLL, and YLD to holistically assess cancer burden and align with global standards [10]. Incidence data highlight trends and risks, while prevalence reflects growing survivorship care needs, especially in low-resource settings [1, 9]. Mortality and survival rates benchmark healthcare performance, revealing disparities in diagnosis and treatment access [1]. YLL and YLD capture cancer’s societal and economic impacts, addressing both mortality and survivorship challenges [2]. This integrated approach surpasses the limited focus of many global systems, such as those by Wei et al. [34]. For instance, the Global Burden of Disease Study offers detailed estimates of cancer metrics for global impact assessment [18], and the U.S. Cancer Statistics Data Visualization Tool provides granular insights into incidence, mortality, and survival, supporting public health initiatives [16].

The CSS incorporates advanced geographic filters operating at multiple spatial resolutions, primarily at the provincial and county levels, to overcome the limitations of broad-scale analyses and facilitate nuanced interpretations of cancer patterns. As established in spatial epidemiology literature, varying geographic resolutions can either reveal or mask significant spatial heterogeneity; for example, provincial-level analyses may conceal sub-regional disparities, whereas county-level granularity can identify localized hotspots driven by environmental, demographic, or infrastructural factors. The CSS enables users to switch between these resolutions, promoting an understanding of how spatial aggregation influences observed trends, broader provincial views support policy-level decision-making, while county-level insights enable targeted interventions. This multi-resolution approach aligns with principles of spatial justice and precision public health, ensuring equitable representation across communities. Furthermore, the integration of spatial autocorrelation and hotspot analysis tools allows users to statistically validate the consistency of patterns across resolutions, distinguishing true trends from scale-related artifacts. Through these capabilities, the CSS not only visualizes but also interprets spatial complexity, empowering stakeholders to make informed, context-sensitive decisions.

This aligns with Goovaerts and Dowell, who highlight the importance of geographic granularity in addressing health inequities [35, 36]. The integration of advanced spatial analysis tools further enhances the identification of high-risk areas, facilitating targeted interventions [7]. Spatial modeling of cancer survival has proven critical in identifying geographic disparities and informing resource allocation. The growing use of spatially referenced data in cancer studies reflects advancements in spatial analytics and recognition of geographic health influences [37, 38]. Fayet emphasized the importance of spatial dimensions in uncovering health disparities and guiding regional interventions [38], while Baade demonstrated the utility of spatial survival methods in optimizing resource distribution [37]. These findings underscore the global relevance of incorporating fine-grained spatial analysis into cancer surveillance systems.

A key innovation in this study was the incorporation of multiple standard populations (e.g., SEGI, WHO, and Iranian-specific populations) for age standardization, improving cross-regional comparisons and minimizing demographic biases [39]. This adaptability aligns with IACR recommendations and enhances the reliability of local and global health assessments. Corazziari highlighted the importance of using multiple standards to improve comparability across studies [40]. WHO’s new World Standard Population further supports precise age standardization for international comparisons [39], while SEER provides diverse standard populations to aid researchers in age-adjusting cancer rates [41]. These advancements highlight the critical role of multiple standard populations in ensuring accurate and comparable cancer surveillance.

The use of ICD-O coding enhances data consistency and granularity, enabling detailed epidemiological insights and effective tracking of underrepresented cancers. For example, distinguishing between stomach cancer (C16) and Cardia cancer (C16.0) supports targeted public health interventions [42]. This precision advances personalized medicine and strengthens correlations between cancer subtypes and genomic data [43]. ICD-O coding also aids in identifying and classifying rare cancers, improving epidemiological data accuracy. The SEER program’s site recode classifications based on ICD-O-3 standardize cancer reporting [41], while WHO’s ICD-O-3 guidelines ensure uniformity across registries [44]. These advancements enhance the analysis of cancer incidence and outcomes.

Phase two: system design and development

The CSS developed in this study represents a significant advancement in cancer data management, addressing critical challenges in public health monitoring and decision-making. Tailored for Iran, this integrated platform provides on-demand monitoring, spatial analysis, and predictive modeling, surpassing existing systems with its innovative features and adaptability.

A key innovation is its on-demand monitoring capability, distinguishing it from platforms like GCO that offer only annual updates [10]. On-demand access enables immediate trend detection, facilitating timely interventions, aligning with Khoury et al.‘s findings on the importance of dynamic systems for chronic disease management [45]. The system’s trend analysis further enhances temporal evaluations with detailed filters for age, gender, location, and cancer type, offering insights often absent in traditional platforms. Predictive capabilities for 5-, 10-, and 20-year horizons support strategic planning, reflecting Mistry et al.‘s emphasis on long-term projections as essential public health tools [46].

The integration of advanced spatial analysis tools is a key achievement, incorporating features such as spatial autocorrelation, hotspot analysis, and cluster identification—previously restricted to GIS-specific platforms like ArcGIS. These tools enable precise identification of high-risk areas and resource optimization, aligning with Roquette et al.’s findings on the value of spatial analysis for understanding geographic disparities [47]. Robertson et al. further highlighted the importance of combining temporal and spatial analyses, a capability effectively implemented in this system [48]. The system’s predictive and risk factor analytics position it at the forefront of health informatics, leveraging AI’s transformative potential, as noted by Topol [49]. The relative risk analysis module provides insights into causal relationships, supporting targeted interventions, as emphasized by Rothman et al. [50]. On-demand multivariate analyses enable the exploration of demographic, environmental, and infrastructural determinants, uncovering complex interactions. Brenner et al. underscore the value of such approaches for tailoring interventions and optimizing resource allocation [51]. Lansdorp et al. demonstrated that advanced surveillance systems could reduce preventable cancers by 15%, illustrating the transformative impact of data-driven insights [49].

Managing the CSS’s extensive dataset, including over 400 topological codes, 56,000 morphological codes, and 20 million records, required strategic optimization. Pre-computed tables were implemented to store metrics for specific filter combinations, reducing computational load and enabling rapid data retrieval. Tables were organized across six dimensions—provincial, county, year, gender, topological, and morphological—to support granular filtering and precise analyses, ensuring efficient data handling critical for public health planning (Fig. S38). From a technical perspective, the adoption of pre-computed tables ensures system efficiency and scalability. By storing calculated values for common filter combinations, the system minimizes computational burden during peak usage, providing rapid and accurate responses. This approach not only enhances performance but also supports the handling of large datasets, as highlighted in comparisons with SEER and other global systems.

The CSS employs a modular architecture developed with Django and Vue.js frameworks, ensuring flexibility and maintainability, which aligns with findings by Agarwal et al. [52], that underscore the efficacy of modern development frameworks in health informatics applications. While alternative backend frameworks such as Flask offer flexibility for lightweight applications, they necessitate additional configuration to achieve comparable functionality, potentially extending development timelines; future research may explore these alternatives to identify potential enhancements. Similarly, for frontend development, frameworks like React or Angular, despite their robust ecosystems and rendering efficiencies, introduce greater complexity that could impede development progress; their potential benefits merit further investigation in subsequent studies to optimize frontend performance in health informatics systems. Comparatively, this study advances the field of cancer surveillance by addressing gaps identified in previous systems. For example, Baade et al. noted the limitations of static, annual data in existing platforms, which restricts timely decision-making [37]. This study overcomes such barriers by offering on-demand functionality, granular geographic filters, and multivariate risk factor analysis, setting a new standard for comprehensive cancer surveillance systems.

Phase three: system evaluation

The system evaluation utilized Nielsen’s Heuristic Usability Assessment, a well-validated methodology recognized for its efficiency in identifying usability issues with high reliability, even with a limited panel of experts [30, 53, 54]. Given the developmental stage and constraints related to funding and infrastructure, this approach was selected as a cost-effective means to assess functionality, user experience, and overall usability, focusing on structured feedback from domain experts, including medical informatics specialists, pathologists, and public health managers. These evaluations yielded critical insights into interface design, navigational clarity, error management, and functional consistency, key determinants of long-term usability and adoption in clinical and administrative contexts. While external validation through field testing or pilot implementation would further strengthen the system’s generalizability, the current heuristic evaluation findings establish a solid basis for future iterative refinement and deployment planning. We recommend initiating localized pilot studies in selected provincial cancer centers to assess real-world applicability, user acceptance, and seamless integration into existing workflows. Its user-friendly interface supports accessibility for diverse users, including those with limited technical expertise, corroborating Kushniruk et al.‘s emphasis on usability in optimizing health informatics systems [54]. The intuitive design enhances navigation, echoing Zahabi et al.‘s findings on the role of visual interfaces in improving healthcare workflows [53]. The system’s user-centered approach addresses the diverse needs of stakeholders, as highlighted by Bauer et al., who noted the importance of such designs in fostering adoption across public health and clinical domains [55]. By simplifying data analysis and ensuring functionality for users with varying expertise, the CSS demonstrates significant utility for policymakers, researchers, and healthcare providers. This evaluation reinforces the transformative potential of advanced health information systems in enhancing cancer surveillance and public health management.

Public health implications

The CSS developed in this study establishes a robust framework for enhancing cancer monitoring and control, specifically tailored to address Iran’s public health needs while offering scalability for countries with comparable healthcare challenges. Its modular design integrates GIS-based spatial analysis, predictive analytics, and multi-level data standardization to deliver granular, on-demand insights into cancer trends and risk factors. These capabilities enable evidence-based policymaking, targeted interventions, and equitable resource allocation by prioritizing high-risk populations and reducing late-stage diagnoses. Predictive models forecast cancer trends over 5-, 10-, and 20-year horizons, supporting proactive planning and prevention strategies. Additionally, GIS functionalities identify spatial disparities, ensuring resources are directed to underserved regions, aligning with World Health Organization strategies for equitable cancer control through early detection and improved health outcomes. The CSS is designed for scalability and interoperability, adhering to structured data modeling principles that facilitate future integration with international standards such as HL7 FHIR and SNOMED CT. Although integration with these standards was not feasible at this stage, its RESTful API and use of standardized coding (e.g., ICD-O-3) enable potential alignment with national electronic health record systems, regional cancer networks, and global health observatories. This positions the CSS to contribute to harmonized cancer surveillance across jurisdictions, fostering collaborative research and globally interconnected public health infrastructure.

Limitations and future directions

The CSS exhibits considerable potential, though several limitations warrant acknowledgment to ensure a thorough evaluation of its current implementation. The system is hindered by restricted access to recent cancer registry data and incomplete clinical datasets, which may undermine the timeliness and depth of epidemiological analyses. Reliance on aggregated data from statistical agencies, rather than individual-level data, diminishes analytical granularity, constraining the identification of specific risk associations for localized interventions. The use of static population figures, due to the absence of annual census data, may compromise the accuracy of crude and age-standardized incidence rates. The usability assessment, conducted via Nielsen’s Heuristic Usability Assessment, was limited by cost and infrastructural constraints, precluding external validation or clinical formative testing, thus restricting generalizability to real-world and low-resource contexts. The predictive modeling module is affected by data quality issues, including underreporting, inconsistent coding practices, and limited rural coverage, alongside static demographic data skewing projections; validation was confined to internal checks and historical comparisons, lacking external prospective validation. Ethical concerns in AI-assisted predictions, mitigated by interpretable models, necessitate ongoing user education and human oversight to address transparency and bias risks. Implementation challenges include computational reliance on precomputed tables due to the absence of on-demand retraining, and data privacy, managed through Role-Based Access Control and JWT authentication, though advanced frameworks like federated learning remain unintegrated. Furthermore, alignment with international interoperability standards (e.g., HL7 FHIR, SNOMED CT) was not achieved due to infrastructural limitations, restricting global applicability.

To address these limitations and enhance the CSS’s efficacy, we propose the following prioritized recommendations for future research and development. First, establishing robust collaborations with national health authorities is essential to secure comprehensive, up-to-date cancer registry data and integrate clinical and genomic information, thereby enriching data coverage and epidemiological insights. Second, conducting external validation through pilot implementations in diverse clinical settings, particularly low-resource environments, is critical to evaluate real-world performance, user satisfaction, and adaptability, thereby bolstering global relevance. Third, advancing predictive modeling necessitates integrating machine learning with robust validation frameworks (e.g., sensitivity analyses, uncertainty quantification) and prospective validation, alongside enhanced transparency and oversight to mitigate ethical concerns. Fourth, improving scalability and interoperability requires alignment with international standards like HL7 FHIR and SNOMED CT, leveraging the RESTful API for cross-border collaboration. Fifth, addressing computational constraints and data privacy entails adopting cloud-based solutions for on-demand retraining and integrating federated learning with encrypted data lakes.

Conclusion

This study introduces a Cancer Surveillance System designed to address critical gaps in cancer monitoring, prevention, and control in Iran. By integrating on-demand analytics, predictive modeling, and spatial analysis, the system provides comprehensive insights into cancer trends, disparities, and risk factors. The use of multiple standard populations and ICD-O-3 classifications enhances data accuracy and granularity, enabling equitable resource allocation and evidence-based decision-making. Beyond its regional scope, the system establishes a benchmark for modern cancer surveillance, demonstrating how advanced technologies can address limitations of traditional platforms. Its modular design ensures adaptability across diverse healthcare settings, offering a scalable model for nations facing similar public health challenges. The GIS integration facilitates targeted interventions by identifying high-risk areas, while predictive analytics support proactive planning and resource optimization. This system bridges the gap between traditional registries and next-generation surveillance platforms, highlighting the transformative potential of modern CSS in reducing cancer burdens, advancing precision public health, and achieving global health equity. Its findings contribute to global cancer surveillance efforts, providing a replicable framework to improve public health outcomes and address systemic disparities in cancer care.

Abbreviations

CSS

Cancer surveillance system

ICD-O-3

International classification of diseases for oncology, 3rd Edition

GIS

Geographic information system

GCO

Global cancer observatory

UML

Unified modeling language

PRISMA

Preferred reporting items for systematic reviews and meta-analyses

YLD

Years lived with disability

YLL

Years of life lost

WHO

World Health Organization

ECIS

European cancer information system

CVR

Content validity ratio

ZUMS

Zanjan University of Medical Sciences

RBAC

Role-based access control

API

Application programming interface

ORM

Object-relational mapping

SPA

Single-page application

JSON

Javascript object notation

REST

Representational state transfer

JWT

JSON Web Token

IQR

Interquartile range

RR

Relative risk

N

Number

ASR

Age-standardized rate

HH

High-high

LL

Low-low

HL

High-low

LH

Low-high

OLS

Ordinary least squares

Authors’ contributions

M.S., as the lead researcher and primary author, conceptualized and designed the study, executed data collection and analysis, designed and developed the Cancer Surveillance System, and drafted and revised the manuscript. M.G.S., as the project supervisor, provided critical guidance and contributed to manuscript revisions. S.M.A. and A.J. offered valuable insights to enhance the research and manuscript quality. All authors collaborated to ensure methodological rigor and analytical accuracy, endorsing its content and conclusions for publication.

Funding

This study did not receive any funding or financial support.

Data availability

All datasets analyzed during this study are derived from publicly available sources and are included in this article and its supplementary materials. No proprietary, restricted, or confidential datasets were used in the study.

Declarations

Ethics approval and consent to participate

This investigation was undertaken as a component of a Ph.D. thesis in Medical Informatics by M.S., with ethical approval granted by the Tehran University of Medical Sciences under protocol number IR.TUMS.SPH.REC.1401.260. The study complies with the ethical guidelines set forth by the institutional research committee and its subsequent revisions. Given that the research entailed the synthesis and analysis of publicly accessible data, formal participant consent was deemed unnecessary

Consent for publication

Not applicable

Competing interests

The authors declare no competing interests.