Eyecare-cloud: an innovative electronic medical record cloud platform for pediatric research and clinical care

Abstract

Background and objectives

Clinical data are essential for developing cloud platforms for intelligent diagnosis and treatment decision of diseases. However, cloud platforms for data sharing and exchange with clinicians are poorly suited. We aim to establish Eyecare-cloud, a platform which provide a novel method for clinical data and medical image sharing, to provide a convenient tool for clinicians.

Methods

In this study, we displayed the main functions of Eyecare-cloud that we established. Based on clinical data from the cloud platform, we analyzed the incidence trend of the most common infantile retinal diseases, such as retinopathy of prematurity (ROP), over the past 20 years, as well as the associated risk factors for ROP occurrence. Statistical analyses were performed using GraphPad Prism (V.8.0) and SPSS software (V.26.0).

Results

The Eyecare-cloud offers numerous advantages, including systematic archiving of patient information, one-click export data, simplifying data collection and management, eliminating the need for manual input of clinical information, reducing clinical data migration time, and lowering data management costs significantly. A total of 22,913 premature infants from Eyecare-cloud were included in the data analysis. Based on 20 years of premature infant screening data analysis, we found that the ROP incidence began to slowly decline starting in 2003 but showed a gradual increase trend again in 2016. The incidence of severe ROP remained relatively stable at a low level since 2010. The number of premature infants increased steadily before 2016 but decreased since then. ROP occurrence was significantly associated with male sex, lower gestational age, and lower birth weight (P < 0.001).

Conclusion

Eyecare-cloud provides clinicians and researchers with convenient tools for big data analysis, which helps alleviate clinical workloads and integrate research data. This cloud platform supports the principles of predictive, preventive, and personalized medicine (PPPM/3PM), empowering clinicians and researchers to deliver more precise, proactive, and patient-centered eye care.

Introduction

With the worldwide digitalization of medical records, efficient management of massive healthcare data, including electronic health records, examinations, and outpatient follow-up, becomes crucial [1]. The potential of electronic records to enhance healthcare quality and efficiency [2] brings convenience to both clinical work and research [3, 4]. Faced with massive medical data, innovative solutions are urgently needed to drive research innovation and elevate the standard of clinical services.

Aggregating data enhanced with real-world clinical settings can provide point-of-care evidence to doctors. Efficient clinical practice depends on effective information management, access, use, and reuse of data. However, the inflexibility of most platforms stands as a significant barrier to data linkage and scaling across healthcare systems [5]. To ensure efficient data utilization, addressing concerns of interoperability, standardization, access, and real-time analysis is crucial [6]. Sufficient computing power and algorithms are widely available [5], necessitating exploration of ways to integrate them into our clinical works. Technology enables closer contact with patients beyond clinical settings [5]. A timely clinical intervention is conducive to enhancing the vision of high-risk patients with retinopathy of prematurity (ROP) [7]. As emphasized in the European Association for Predictive, Preventive and personalized medicine (EPMA) white paper, identifying ROP is crucial to preventing blindness due to ROP [8]. The implementation of a cloud platform can help prevent missed examinations, follow-ups, and medical errors [9].

Working hypothesis and study aims in the PPPM framework

With the promotion of neonatal eye disease examinations and the development of retinal imaging devices, many maternal and child health hospitals have purchased relevant equipment. This has enabled non-ophthalmic doctors to independently conduct eye disease screenings. However, these doctors’ diagnostic proficiency for infantile retinal diseases has not kept up with the advancements. This triggered various changes in the delivery of healthcare, including a sudden and rapid surge in the number of telemedicine visits. Telemedicine has resulted in an increased time commitment from doctors for the writing of electronic reports. Consequently, we need to explore novel strategies to alleviate these clinical burdens [10]. There are already some electronic cloud platforms available [11, 12]. Physicians reported positive effects of these systems on several dimensions of quality of care and high levels of satisfaction [13]. However, there is relatively little research on cloud platforms specifically related to the diagnosis and treatment of pediatric fundus diseases.

We developed an integrated clinical platform for pediatric fundus diseases called Eyecare-cloud. This platform facilitates the intelligent collection of research medical records, extraction, integration, and consolidation of disease-related data, standardized datasets, and the provision of normalized electronic medical record reports and a follow-up system. Therefore, the Eyecare-cloud supports research in clinical diagnosis, treatment, and prognosis, contributing to the enhancement of the quality of clinical studies and the improvement of diagnostic and treatment capabilities. This study contributes to the intelligent diagnosis, treatment, progression, and prognosis assessment of pediatric fundus diseases. Therefore, it aligns with the concept of PPPM.

Methods

This is a retrospective and hospital-based study that adhered to the tenets of the Helsinki Declaration and was approved by the Medical Ethics Committee of Shenzhen Eye Hospital (SZEH, 2023KYPJ091). Parents of the infants provided written consent before the ophthalmic examination at SZEH. All images and data de-identification were done before analysis.

Cloud platform application in clinical settings

The initial version of the cloud platform can be considered a primary electronic medical management system, requiring manual input of all patient information to create electronic records [14]. During the 20 years of development, we continuously updated the system performance. In April 2020, the new version with numerous functions was successfully launched, and the current platform provides more convenient medical services (Fig. 1).

Fig. 1.

The integration of Eyecare-cloud with clinical workflow and research. SLO, scan laser ophthalmoscope; OCT (A), optical coherence tomography (angiography)

This platform mainly consists of three sections: the nursing section, the doctor section, and the super administrator section. The nurse section is to check the accuracy of baseline information and medical history records of infants filled out by patients’ family members through the WeChat official account and uploaded to our platform. For patients who are not familiar with WeChat appointment procedures, nurses can assist them with appointments through a nursing interface. After the appointment is completed, the relevant informed consent form will also be automatically generated and printed by nurses. Simultaneously, patient information will be synchronized to related doctor’s and the super administrator’s interface.

For infants under 1 year old, after the pupil dilation and topical anesthesia procedures, the fundus image capture will be initiated. During the image capture phase, we just need to input the infant’s ID number on the machine to search for their information. Then, simply click the capture button to implement the examination directly. The image capture system included various fundus imaging systems (RetCam III, USA; SW-8000, China; RS-B002, China; et cetera). The Eyecare-cloud application can save us time in establishing baseline information for new patients during image capture, as this information is already filled out by the patients’ families through the WeChat official account at the onset and automatically synchronized with the imaging device. This is a functionality that other hospitals not using this cloud platform do not have. After the capture is completed, all examination data will be automatically synchronized to the cloud platform in real-time for doctors to generate electronic reports.

For infants over 1 year old, we can provide corresponding examinations based on their degree of cooperation, including pupil dilation refraction, ultra-widefield (UWF) imaging, optical coherence tomography (OCT) and OCT angiography (OCTA) imaging, corneal topography, and IOLMaster 700 examination. The platform stands out from traditional tools by seamlessly linking with third-party data sources, including hospital information system (HIS) and picture archiving and communication system (PACS) [12], achieving mutual communication and connectivity. All examination results in clinical settings sent to the HIS are automatically transmitted to Eyecare-cloud through PACS as well, including uncorrected visual acuity, refractive results, intraocular pressure (IOP), OCT (A) reports, UWF images, corneal topography reports, and IOL master reports (Fig. 1). We can view various examination results on the cloud platform once the examinations are completed. Figure 1 displays the integration of the cloud platform with clinical workflow and research. By clicking on the edit button, we can enter the report writing section, complete, and print electronic reports in real-time to patients. The platform displays all follow-up and treatment medical records, enabling doctors to review them at their convenience. The electronic reports completed by doctors are also synchronized in real-time to the respective WeChat official account’s personal medical records of the patients’ families.

The super administrator section incorporates capabilities such as patient information editing and management, cohort studies registration, multimodal data collection, biobank information, administrator account management, data security management, data retrieval, and other functionalities. The electronic medical record in this cloud platform complies with national standards for industry-specific norms and international disease diagnosis and classifications.

Cloud platform-related data collection

The initial purpose of establishing this cloud platform was for ROP infant management. During 20 years of clinical experience, we observed that, besides ROP, there are numerous other pediatric fundus diseases requiring diagnosis and treatment. Consequently, the platform has been expanded to encompass various pediatric fundus diseases, including familial exudative vitreoretinopathy (FEVR), coats disease, retinoblastoma (RB), retinitis pigmentosa (RP), coloboma of the choroid, and congenital retinal fold (CRF). The platform consisted of all clinical information of pediatric patients, including baseline characteristic information, treatment-related information (specific treatment modalities and dates), historical medical records, ophthalmic examination results, fundus photographs, and generating electronic reports, among other details.

We have established cohorts for ROP and FEVR on this platform. When enrolled patients in these cohorts for follow-up, the platform will indicate their cohort membership, assisting doctors in completing comprehensive follow-ups for patients. The platform has set up a reminder function, which will send timely follow-up reminders to the families of children based on the follow-up dates indicated in the electronic reports. We have also opened a follow-up channel during the school-age children’s winter and summer vacations, reminding families to bring their children for timely follow-ups. For children who have not come for follow-up for an extended period, we will also make phone calls to remind them.

Inclusion and exclusion criteria of ROP-related data

Considering the extensive dataset of premature infants on the cloud platform, we opted to focus our subsequent analysis specifically on data related to premature infants. From January 2003 to September 2023, the platform has recorded a total of 41,066 cases. We conducted an analysis on the cases of 22,913 premature infants for whom complete data were available. Patients diagnosed with other diseases or with incomplete data were excluded (Fig. 2).

Fig. 2.

Flow chart of participants and examinations in this study. UWF, ultra-widefield; OCT (A), optical coherence tomography (angiography)

Statistical analyses

Statistical analyses were conducted using GraphPad Prism (V.8.0) (GraphPad Software Inc.) and SPSS software (V.26.0) (IBM). Student’s t-test was used for comparison of continuous variables, while the chi-square test was used for comparison of categorical variables. The Mann–Whitney test was performed to evaluate the differences between the two groups. Univariate and multiple logistic regression models were used to investigate the associations between incident ROP and risk factors including gender, gestational age, and birth weight. Statistical significance was set at P < 0.05.

Results

The clinical application of the cloud platform

Currently, the Eyecare-cloud has been used in clinical settings in many areas. More than 40 hospitals are using Eyecare-cloud for electronic medical report writing and telemedicine. This establishes a robust data foundation for multicenter real-world study. The medical reports are shown in Figs. 3 and 4. Figure 3a shows the electronic report of an ROP patient requiring laser therapy, while Fig. 3b shows the report of a patient with spontaneously regressed ROP. Figure 4a shows the electronic report of an ROP patient requiring anti-vascular endothelial growth factor (VEGF) agent treatment, while Fig. 4b shows a report of a patient with retinoblastoma containing treatment suggestions.

Fig. 3.

Examples of electronic reports for infants with treatment-requiring ROP and spontaneously regressed ROP. ROP, retinopathy of prematurity

Fig. 4.

Examples of electronic reports for infants with treatment-requiring ROP and retinoblastoma. ROP, retinopathy of prematurity

Demographics of the premature infants

A total of 22,913 premature infants were included in this data analysis. Infants born through eutocia and cesarean section were 8895 and 14,018, respectively. Single births totaled 15,885, with 7028 being multiple births. Males and females numbered 12,855 and 10,058, respectively (Table 1).

Table 1.

Baseline characteristics of premature infants

Item	Non-ROP	ROP	Total	OR	P value
Birth modes
Eutocia	6969	1926	8895	0.467	< 0.0001
Cesarean	12,415	1603	14,018
Pregnancy
Singleton	13,407	2478	15,885	0.951	0.21
Multiparity	5977	1051	7028
Gender
Male	10,684	2171	12,855	0.768	< 0.0001
Female	8700	1358	10,058
Birth weight	1528 ± 325	1189 ± 341	/	/	< 0.0001
Gestational age	31.80 ± 2.29	28.96 ± 2.41	/	/	< 0.0001
Total	19,384	3529	22,913	/	/

Incidence of ROP and treatment-requiring ROP

Based on 20 years of premature infant screening data analysis, we found that the incidence of ROP began to slowly decline starting in 2003 but showed a gradual increase trend again in 2016. The incidence of severe ROP remained relatively stable at a low level since 2010. The number of premature infants increased steadily before 2016 but decreased since then. The gestational age and birth weight of preterm infants have shown a gradual decrease trend (Table 2, Fig. 5).

Table 2.

Annual incidence of ROP and treatment-requiring ROP

Year	Premature infants	ROP	Treatment-requiring ROP
2003	9	2 (22.22)	1 (11.11)
2004	34	20 (58.82)	9 (26.47)
2005	153	34 (22.22)	21 (13.73)
2006	199	52 (26.13)	25 (12.56)
2007	373	75 (20.11)	37 (9.92)
2008	573	131 (22.86)	66 (11.52)
2009	669	137 (20.48)	71 (10.61)
2010	1176	168 (14.29)	60 (5.10)
2011	1130	165 (14.60)	47 (4.16)
2012	1459	200 (13.71)	67 (4.59)
2013	1338	182 (13.60)	58 (4.33)
2014	1695	196 (11.56)	88 (5.19)
2015	1733	221 (12.75)	106 (6.12)
2016	2383	230 (9.65)	129 (5.41)
2017	2046	309 (15.10)	115 (5.62)
2018	1584	288 (18.18)	113 (7.13)
2019	1574	287 (18.23)	91 (5.78)
2020	1352	199 (14.72)	56 (4.14)
2021	1104	209 (18.93)	71 (6.43)
2022	1342	263 (19.60)	74 (5.51)
2023	987	161 (16.31)	36 (3.65)
Total	22,913	3529 (15.40)	1341 (5.85)

Fig. 5.

Trends of baseline information and ROP incidence over 20 years. ROP, retinopathy of prematurity; TR-ROP, treatment-requiring ROP; GA, gestational age; BW, birth weight

Logistic regression analysis of risk factors

In univariate and multiple logistic regression models, ROP occurrence was significantly associated with male sex, lower gestational age, and lower birth weight (P < 0.001) (Table 3).

Table 3.

Risk factors associated with ROP occurrence

Risk factors	Univariate		Multiple
	OR	P value	OR	P value
Gender	0.768 (0.714, 0.827)	< 0.001	0.797 (0.733, 0.867)	< 0.001
Gestational age	0.606 (0.595, 0.617)	< 0.001	0.689 (0.672, 0.705)	< 0.001
Birth weight	0.997 (0.997, 0.997)	< 0.001	0.999 (0.999, 0.999)	< 0.001

Big data export and retrieval function

The cloud platform has stored information for over 41,066 pediatric cases, including their various baseline information and examination results. Doctors and researchers can retrieve and export required data based on various criteria. For example, they can export all baseline information and examination results of children diagnosed with ROP, or export information and examination results of all children who received anti-VEGF treatment. Exported data can be presented in Excel form or as fundus images from eye examinations, imaging examination reports, et cetera, according to the requirements of doctors or researchers. We have now exported over 600,000 fundus images of all infants in the Eyecare-cloud, which will be used for innovative research in intelligent diagnostic and therapeutic technologies.

Cohort follow up

The ongoing cohort studies using this platform include the following: follow-up cohort for patients with FEVR (241 patients), cohort for anti-VEGF treatment (233), and laser therapy (1021) of ROP, et cetera. Cohort-enrolled patients will undergo routine follow-ups every 6 months to observe the disease progression.

Data privacy and security

All data access requires a personal account. Different doctors logging into the platform can only access information about patients they received for diagnosis and treatment. If there is a need to access all information, higher-level permissions are required. Only by logging in with the highest-level access account can doctors or researchers obtain all the data.

Discussion

Eyecare-cloud serves as a convenient tool for clinicians and researchers, accelerating the technical aspects of clinical works, and enhancing both the quality and efficiency of clinical studies. The cloud platform offers numerous advantages, including systematic archiving of patient information, one-click export data, simplifying data collection and management, eliminating the need for manual input of all information to provide a streamlined process requiring only the checking relevant conditions for each examination, reducing clinical data migration time, and lowering data management costs significantly. Additionally, the platform seamlessly integrates data science, biomedical informatics, telemedicine, and medical imaging, collectively contributing to the advancement and implementation of artificial intelligence (AI).

For researchers, data collection and management are fundamental and crucial. As one of the capabilities of the specialized disease service platform, the Eyecare-cloud provides database construction and intelligent management capabilities. Targeting the clinical research domain, this application employs an efficient and automated research method encompassing database creation, verification, and analysis. Electronic medical records have already created significant value in clinical settings [1, 6, 13, 15–18]. For example, Ghazi. et al. have implemented alert systems through electronic medical records to timely remind clinicians of heart failure-related treatment information [15] or kidney injury-related information. Jillian. et al. used data from standardized electroencephalogram reports to generate prediction models for vulnerable neonates [16]. J Marc. et al. used electronic health records to support the care delivery process is a concern for the US healthcare system [17]. The era of exascale supercomputing, reaching computing power in the order of hundreds of quintillions, is approaching [19]. Beneath this unprecedented computational capacity, we anticipate the realization of various platforms dedicated to clinical services. We should fully leverage the tremendous computing power and apply it to the maximum extent in clinical work. Efforts to enhance the intelligence of electronic medical reports are currently underway in a systematic method.

In recent years, AI has rapidly evolved [5], offering a new approach to neonatal medical management that has the potential to revolutionize ROP prevention [20]. As AI develops further, we will increasingly rely on it as a tool for decision-making [21, 22]. Our platform supports AI development by allowing the retrieval of all relevant images and related data for downloading and training AI models. While technological advancements hold the potential to revolutionize medicine through the comprehensive collection and analysis of clinical data, the progress in clinical integration lags far behind [5]. Therefore, the establishment and application of our platform will offer more insights to facilitate the integration of clinical data. We have successfully implemented automatic identification and analysis of ROP and other fundus diseases using infant fundus images [23–32]. In the future, we anticipate embedding the developed AI models into the report writing section of the cloud platform. This is expected to achieve automated electronic report generation, further reducing the manual tasks of image analysis and report writing for clinicians. This may relieve doctors from burdensome clinical tasks, alleviate their fatigue, and ultimately achieve greater value in healthcare. Doctors will only need to recheck and print the reports for patients, freeing up more time for communication with the parents of the infants and enhancing doctor-patient relationships.

In the current study, based on 20 years of premature infant data analysis obtained from the Eyecare-cloud, we found that the incidence of ROP began to slowly decline starting in 2003 but showed a gradual increase trend again in 2016. It is speculated that the decrease in ROP incidence after 2004 was due to the publication of oxygen therapy guidelines for premature infants in China. However, the slight increase after 2016 might be attributed to better neonatal care, resulting in more premature infants surviving and increasing the incidence of ROP. The number of premature infants increased steadily before 2016 but decreased since then. This trend is likely attributed to the continuous decrease in newborn birth rates in China in recent years.

Eyecare-cloud can also assist in clinical study analysis. We exported data of FEVR patients from the platform and conducted clinical analysis. We found the differences in anterior segment parameters in FEVR patients compared to those of normal children, offering novel assessment perspectives for diagnosing FEVR [33]. In addition, we analyzed the foveal microvascular structure in ROP patients who had undergone anti-VEGF treatment or laser therapy using OCTA. We concluded that anti-VEGF treatment might contribute to a reduction of foveal vessel density, whereas laser therapy might contribute to a smaller foveal avascular zone and thicker foveal thickness [34]. Furthermore, our platform conducted a prospective randomized controlled trial that compared the effectiveness of two types of anti-VEGF agents in treating ROP. The results revealed that a novel Chinese agent (conbercept) demonstrated notable therapeutic efficacy [35]. Recently, we also used the platform to export data on preterm infants and construct an ROP prediction model, further demonstrating the significant value of the platform in clinical research [36]. Based on the clinical data analysis in our study, we found that the occurrence of ROP was significantly associated with male sex, lower gestational age, and lower birth weight. These risk factors have also been confirmed by previous research [37]. We further validated existing research findings through the large dataset collected by the Eyecare-cloud. We look forward to leveraging the Eyecare-cloud to gather a broader range of data, uncover more risk factors associated with ROP occurrence, and provide new insights for the prevention of ROP.

The presentation of standardized training for ophthalmologists in different regions might highlight the advantages of various systems, thereby encouraging the establishment of a more standardized training framework [38]. Similarly, the demonstration of this cloud platform we have established is intended to serve a similar purpose. We hope to assist more pediatric ophthalmologists in achieving the digitization, standardization, and uniformity of clinical works and researches. We also hope to receive suggestions from more professionals worldwide through published articles to further improve the performance of our platform, providing better services for the young generation.

Conclusion, outlook in the framework of PPPM/3PM, and limitations of the study

Conclusion

Eyecare-cloud offers clinicians and researchers with convenient tools for big data analysis, facilitating the provision of electronic medical records. This not only relieves clinical workloads but also enhances the quality and efficiency of clinical studies.

Limitations and outlook in the framework of PPPM/3PM

This study has the following limitations. First, the paper format limits the vividness of the platform’s presentation. However, we have outlined the main advantages and application scenarios of the cloud platform. If readers require further insight, they can engage with us for more in-depth communication. Second, this study has provided a preliminary introduction to the cloud platform and conducted a macroscopic analysis of big data, but deeper exploration has not yet been undertaken. In future research, more in-depth analyses are urgently needed to demonstrate the advantages of the cloud platform in clinical applications and research analyses. Eyecare-cloud can streamline clinical workflows and serve as an essential platform for researchers to collect and analyze clinical data, thereby alleviating clinical workloads and enhancing the efficiency of pediatric fundus disease diagnosis, treatment, prediction, and prevention. Therefore, it also aligns with a crucial aspect of the PPPM/3PM approach emphasized at the EPMA World Congress 2019, which focused on customizing and continuously monitoring patients’ clinical parameters to improve treatment outcomes [39]. To facilitate the future application of this cloud platform in this field, it is strongly recommended to consider the following points:

1. Ensure that the cloud platform accurately records all of the clinical data, including but not limited to basic information, related images, and follow-up details, thus serving as a convenient tool for clinicians in real-world clinical settings.

2. Ensure that the functionality of the cloud platform remains stable and continuously updates and improves according to the needs of real-world clinical applications. Guided by the PPPM/3PM approach, the cloud platform provides comprehensive clinical information and image data for each patient visit through functions such as information storage, retrieval, and export. This serves as a powerful tool for clinicians in report writing and offers researchers a platform for big data management, ultimately achieving better health management for patients.

Abbreviations

ROP

Retinopathy of prematurity

TR-ROP

Treatment-requiring ROP

GA

Gestational age

BW

Birth weight

SZEH

Shenzhen Eye Hospital

OCT

Optical coherence tomography

OCTA

OCT angiography

HIS

Hospital information system

PACS

Picture archiving and communication system

UWF

Ultra-widefield

SLO

Scan laser ophthalmoscope

FEVR

Familial exudative vitreoretinopathy

RB

Retinoblastoma

RP

Retinitis pigmentosa

CRF

Congenital retinal fold

VEGF

Vascular endothelial growth factor

AI

Artificial intelligence

Author contribution

Xinyu Zhao, Guoming Zhang, and Zhenquan Wu were involved in all aspects of the study, including conceptualizing and designing the study, analyzing the data, drafting the initial manuscript, and reviewing and revising the manuscript. Yaling Liu, Honglang Zhang, Yarou Hu, Duo Yuan, Xiayuan Luo, Mianying Zheng, Zhen Yu, and Dahui Ma were involved in data collection, coordinated and supervised data collection, carried out the initial data analyses, and critically reviewed and revised the manuscript. All authors critically reviewed and revised the manuscript.

Funding

This study was supported by the National Natural Science Foundation of China (No. 82271103, 82301269, 82301226), the Sanming Project of Medicine in Shenzhen (No. SZSM202311018), the Guangdong Basic and Applied Basic Research Foundation (No. 2022A1515012326, 22201910240002529), the Shenzhen Medical Research Fund (No. C2301005), the Shenzhen Key Medical Discipline Construction Fund (No. SZXK038), the Shenzhen Fund for Guangdong Provincial High Level Clinical Key Specialties (No. SZGSP014), the Shenzhen Science and Technology R&D Fund Project (No. JCYJ20220530153607015), and the China Ophthalmology New Technology Incubation Project.

Availability of data and material

Further information and requests for resources should be directed to the corresponding author (zhangguoming@sz-eyes.com).

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval

This study was approved by the Ethics Committee of Shenzhen Eye Hospital. All institutions abided by the tenets of the Declaration of Helsinki.

Consent to participate

Not applicable.

Consent for publication

This article has been approved for publication by the authors.

Conflict of interest

The authors declare no competing interests.