Using Data Visualization to Detect Patterns in Whole-person Health Data

Abstract

Data visualization techniques are useful for examining large multidimensional datasets. In this exploratory data analysis study, we applied a visualization pattern detection and testing process to de-identified data to discover patterns in whole-person health for adults 65 and older. Whole-person health examines a person’s environmental, psychosocial, and physical health, as well as their health-related behaviors; and assesses their strengths, challenges, and needs. Strengths are defined as assets and capabilities in the face of short-or long-term stressors. We collected data using a mobile application that delivers a comprehensive whole-person assessment using a simplified version of a standardized instrument, the Omaha System. The visualization pattern detection process is iterative, includes various techniques, and requires visualization literacy. The data visualization techniques applied in this analysis included bubble charts, parallel coordinates line graphs, box plots, and alluvial flow diagrams. We discovered six patterns within the visualizations. We formulated and tested six hypotheses based on these six patterns, and all six hypotheses were supported. Adults 65 and older had more strengths than challenges and more challenges than needs (p<.001). Strengths and challenges were negatively correlated (p<.001). Unexpectedly, a subset of adults 65 and older who had many, but not all, strengths had significantly more needs (p=.04). The use of standardized terminology with its inherent data interrelationships was key to discovering patterns in whole-person health. This methodology may be used in future exploratory data analysis research using new datasets.

Data visualization methods have gained popularity for pattern discovery in large complex datasets (Few, 2012; Tukey, 1977). Tukey’s (1977) exploratory data analysis (EDA) method is a widely accepted iterative approach in which investigators inspect graphical displays to identify trends, patterns, and outliers in the data, and then formulate hypotheses for further testing (Few, 2012; Tukey, 1977). Data visualization is an important skill for clinicians and researchers in this data-rich era. Selecting the data visualization techniques that can be useful, however, can be challenging (Monsen et al., 2019; Tukey, 1977).

Data visualization techniques are diverse and depend on the type of data available (Austin et al., 2021; Monsen, Peterson, et al., 2017). Various techniques can be used for pattern discovery, such as heat maps, bubble charts, parallel coordinate line graphs, box plots, and alluvial flow diagrams. Each technique requires visualization literacy to critically examine and identify patterns in data (Monsen et al., 2019). Similarly, there are many software options for developing visualizations, such as Statistical Analysis Software (SAS), Statistical Package for the Social Sciences (SPSS), R, Tableau, and Excel. Practice and experience are key to developing visualization literacy and proficiency (Tukey, 1977).

In this EDA study, we examined de-identified data to discover patterns in whole-person health, focusing on adults 65 and older. Data were collected using a mobile health application, MyStrengths+MyHealth (MSMH), which is a comprehensive whole-person assessment. The standardized assessment within MSMH is a simplified translation of a rigorous, multidisciplinary health terminology, the Omaha System. The Omaha System has been used for holistic assessments and documentation of care since the 1990’s (Austin et al., 2021; Martin, 2005). The Simplified Omaha System Terms (SOST) has been validated at the 5th grade reading level (Austin et al., 2022). Researchers and clinicians may obtain a free MSMH license through the University of Minnesota Office of Technology and Commercialization. Applying visualization techniques to standardized SOST/Omaha System data enables understanding of whole-person health and healthcare outcomes, based on the inherent data structure (Martin, 2005). Previous research has shown that Omaha System data generated by clinicians and/or community members are well-suited for use in visualization studies (Lee et al., 2015; Monsen, Peterson, et al., 2017; Monsen et al., 2018, 2021; Radhakrishnan et al., 2016).

We selected older adults as our focus because by 2060, 23.5% of the U.S. population will consist of adults 65 years and older (Administration on Aging, 2021; US Department of Health and Human Services (HHS), 2021). Approximately 80% of older adults have one or two chronic conditions (Administration on Aging, 2021). With the expected increase in the aging population and the high prevalence of chronic diseases, approaches such as whole-person assessments are needed to support overall health and well-being and to assist in managing chronic conditions (Austin et al., 2021; Monsen, Vanderboom, et al., 2017).

Whole-person health includes a comprehensive view of health: the environment in which a person lives as well as their psychosocial health, physical health, and health-related behaviors (Carter et al., 2015; Zollman et al., 2017). Whole-person health has three components: strengths, health challenges (signs/symptoms), and health needs (types of care). Strengths are defined as assets, skills, capacities, actions, talents, potential, and gifts for each family member, each team member, the family as a whole, and the community (Miles et al., 2006; Rotegard et al., 2010). Strengths can be leveraged to facilitate self-management and support managing emotional and physical health challenges and needs (Chmitorz et al., 2018). Our long-term goal is to improve understanding of whole-person health across the lifespan. Achieving this goal will help to create a needed paradigm shift in healthcare and ultimately facilitate a strengths-based approach to care. The purpose of this EDA study was to apply data visualization techniques to multi-dimensional strengths, challenges, and needs data of adults 65 and older for pattern discovery, hypothesis formulation, and testing to characterize whole-person health.

Methods

The University Institutional Review Board deemed this EDA study of existing de-identified data exempt from review. Data collection for the study occurred using MSMH on electronic tablets with members of the public (N=383) at a large Midwestern state fair in 2017. Respondents self-selected into the study and were given a drawstring backpack ($1.79 value) for completing the MSMH assessment. The time to complete the survey was approximately 15 minutes. Inclusion criteria were being able to speak English and age over 18 years. Minimal demographic data were collected: age by group (18–24; 25–44, 45–64; and 65+) and self-identified gender (Male; Female; and prefer not to answer). Of the data from the 383 adults, we focused on a cohort of adults 65 and older (n=84) for analysis of whole-person health patterns.

Instrument: The Omaha System

The Omaha System is a research-based, comprehensive, standardized terminology. It originated in community care settings, and has been shown to describe all of health and healthcare (Kang et al., 2022; Martin, 2005; Monsen, Schenk, et al., 2015). It consists of three inter-related standardized instruments with documented psychometric properties: the Problem Classification Scheme (patient assessment/problem list), the Problem Rating Scale for Outcomes (problem evaluation), and Intervention Scheme (used for care planning and services) (Martin, 2005). In this study we used the Simplified Omaha System Terms (SOST) version of the Omaha System.

The Problem Classification Scheme is a comprehensive classification used to describe health. There are 42 non-overlapping concepts within the four domains of health: Environmental (My Living), Psychosocial (My Mind and Networks), Physiological (My Body), and Health-related behaviors (My Self-care). Each of the 42 concepts is defined and has a unique set of signs/symptoms. In SOST, strengths are defined as concepts with a status score of “4” (minimal signs/symptoms) or “5” (no signs/symptoms). Challenges are the unique signs/symptoms associated with a specific concept and in SOST are defined as “yes”, “no”, or none apply. In SOST, the “needs” are related to the Invention scheme for each concept. , .and describe problem-specific actions and activities in four categories: Teaching, Guidance, and Counseling (Info/Guidance); Treatments and Procedures (Hands-On Care); Case Management (Care Coordination); Surveillance (Check-Ins).; and N/A if no needs are applicable

Visualization and Pattern Testing

In this EDA study, we used iterative, multi-step visualization methods (Monsen, Peterson, et al., 2017; Tukey, 1977). Steps included visualization development, pattern detection, hypothesis formulation, and testing.

Data visualization techniques

As a general rule of thumb, several iterations are needed to detect patterns in visualizations (Tukey, 1977). Patterns may be detected from multiple perspectives, such as using raw data or computed variables, with data organized by problem concepts, groups, or domains. All visualizations in this study included one or more computed variables. The investigators visualized strengths, challenges, and needs by group, in various combinations. Visualization techniques included heat maps (Few, 2017; Kirk, 2016), bubble charts (Monsen et al., 2019), parallel coordinate line graphs (Inselberg & Dimsdale, 2009), box plots (Schmidt, 2020), and alluvial diagrams (Dzemyda et al., 2013; Schmidt, 2020).

Heat maps.

Heat maps enable investigators to visualize an entire dataset of raw data, parts of datasets, and/or computed variables or tables to show the relative proportions between values within a selected range of continuous or categorical numeric data (Few, 2017; Monsen, Peterson, et al., 2015). Heat maps may be used as a first step in grossly visualizing the data to detect patterns for further exploration. Heat maps can be created using a spreadsheet application. For example, if using Excel, select the desired data; then apply conditional formatting to rows, columns, single rows, columns, or the entire dataset or table (Kirk, 2016; Monsen et al., 2019).

Bubble charts.

Bubble charts compare data for three variables that may be continuous and/or categorical (Monsen et al., 2019). In a previous study, we developed a three-tiered bubble chart to depict whole-person health for two or more groups using SOST data (Austin, Mathiason, et al., 2022). Bubble charts enable examination of patterns in the distribution of whole-person health data by domain and group (Monsen et al., 2019). Bubble charts were created using a spreadsheet program. If using Excel, data should be arranged according to x-axis, y-axis, and bubble size; then select the data, and insert “Scatter” plot.

Parallel coordinates line graphs.

A parallel coordinates line graph is defined as a series of individual observations of a set of numerical values (Inselberg & Dimsdale, 2009). We adopted parallel coordinate line graphs to observe patterns in strengths, challenges, and needs by problem for a single group (adults 65 and older). Problem concepts are displayed on the x-axis and may be sorted by the Omaha System order or by descending order of strengths, challenges, or needs (Inselberg & Dimsdale, 2009; Schmidt, 2020). If using Excel, follow the steps for multi-series line graphs.

Box plots.

Box plots graphically demonstrate the distribution of the sample and show outliers (Schmidt, 2020). Box plots were generated as a routine component of statistical testing using SAS version 9.

Alluvial (Sankey) flow diagrams.

Alluvial (Sankey) flow diagrams were originally developed to represent the relationships between variables over time (Caban & Gotz, 2015; Schmidt, 2020). In this study, we used an alluvial flow diagram to examine interrelationships among totals of domains for which adults had at least one strength, challenge, and need. We created an alluvial flow diagram using SAS version 9.

Pattern detection

Pattern detection relies on human intelligence to recognize regularities, such as orderliness, symmetry, and consistency, within a data display. Pattern detection is an aspect of human intelligence that arises from the cerebral cortex and includes aspects of visual perception and pre-attentive reasoning to analyze and interpret data (Conway, 2009; Few, 2009). Visualization literacy develops with practice and is an important skill for researchers and clinicians engaged in data interpretation using images (Tukey, 1977).

Hypothesis formulation and testing

Researchers should formulate hypotheses based on patterns using critical analysis and clinical knowledge (Tukey, 1977). The hypotheses should be tested using appropriate statistical techniques based on the hypothesis and the type of data. For this analysis, these patterns, methods, hypotheses, and test results are presented in Table 1. Hypothesis testing using the same data from which the visualizations are created is an integral part of the multi-step iterative pattern discovery process suggested by visualization experts (Berinato, 2016; Tukey, 1977). Supported hypotheses confirm the existence of regularity within random data; these should be tested in other datasets to increase confidence that patterns persist across datasets (Zgraggen et al., 2018).

Table 1.

Average Percent Strengths, Challenges, and Needs Adults 65+ and Adults 18–64

Concept	Strengths		Challenges		Needs
	Adults 65+	Adults 18–64	Adults 65+	Adults 18–64	Adults 65+	Adults 18–64
My Living
Income	63.1	53.8	7.1	26.8	7.1	12.7
Cleaning	59.5	59.9	9.5	13.0	3.6	6.0
Home	64.3	67.9	23.8	20.7	9.5	6.4
Safe at work home	69.0	63.5	10.7	15.1	8.3	5.4
My Mind and Networks
Connecting	60.7	56.5	3.6	7.4	1.2	5.4
Socializing	60.7	50.2	9.5	21.1	2.4	6.4
Role change	39.3	36.8	27.4	28.8	4.8	7.0
Relationships	56.0	51.8	19.0	26.4	4.8	6.4
Spirituality or faith	54.8	41.5	15.5	20.7	2.4	3.7
Grief or loss	52.4	43.8	13.1	16.4	3.6	5.7
Emotions	51.2	43.8	25.0	60.2	2.4	13.7
Sexuality	46.4	49.8	15.5	18.7	3.6	4.3
Caretaking	51.2	47.2	23.8	17.7	8.3	8.4
Neglect	45.2	42.5	4.8	4.3	0.0	3.0
Abuse	46.4	41.8	4.8	7.0	1.2	3.3
Growth and development	59.5	48.8	4.8	9.4	1.2	1.3
My Body
Hearing	38.1	40.5	29.8	16.7	6.0	2.7
Vision	36.9	25.1	47.6	60.9	0.0	4.7
Communication	54.8	48.2	2.4	3.0	0.0	1.0
Oral health	41.7	36.8	16.7	29.4	0.0	4.3
Thinking	42.9	40.1	16.7	22.7	2.4	5.7
Pain	51.2	40.8	6.0	9.7	1.2	4.3
Consciousness	45.2	42.1	0.0	1.0	0.0	0.0
Skin	51.2	38.1	8.3	15.7	0.0	3.3
Moving	39.3	39.5	51.2	34.4	3.6	7.4
Breathing	52.4	44.5	9.5	10.0	1.2	2.3
Circulation	32.1	31.4	66.7	46.5	7.1	7.7
Digesting	42.9	34.4	32.1	34.4	9.5	6.7
Bowel function	44.0	41.5	9.5	9.4	2.4	2.0
Kidney/bladder	29.8	39.8	36.9	18.7	7.1	2.7
Reproductive health	42.9	36.5	1.2	8.0	0.0	1.0
Pregnancy	28.6	23.7	0.0	0.0	0.0	0.3
Postpartum	26.2	20.1	0.0	0.3	0.0	0.0
Abuse	48.8	45.5	8.3	5.4	3.6	1.3
My Self-care
Nutrition	41.7	28.1	46.4	57.9	13.1	12.7
Sleeping	26.2	20.7	75.0	93.0	15.5	17.7
Exercising	38.1	29.4	38.1	38.8	4.8	6.7
Personal care	59.5	49.2	1.2	6.7	0.0	2.0
Substance use	47.6	36.8	1.2	12.7	0.0	2.0
Family planning	39.3	35.5	0.0	2.3	0.0	0.7
Health care	45.2	31.4	22.6	32.8	1.2	3.7
Medications	57.1	41.1	3.6	8.0	0.0	1.7

Results

Of the total sample (N=383), there were 84 adults 65 and older; 299 were aged 18–64 years. In both groups, the majority were female (61.6% and 59.0%, respectively). There were no differences in reported gender between the groups (X²=.833, p>.05). Adults 65 and older averaged 19.8 ± 14.4 strengths. The most frequently reported strengths were for the My Living and My Mind and Networks domain concepts Safe at home and work (69%), Home (64.3%), Income (63.1%), Socializing (60.7%), Connecting (60.7%), and Cleaning (59.5%). Adults 65 and older averaged of 7.4 ± 7.9 challenges. Their most frequently reported challenges were for My Body and My Self-Care domains: Sleeping (75%), Circulation (66.7%), Moving (51.2%), Vision (47.6%), Nutrition (46.4%), Exercising (38.1%), and Kidneys/bladder (36.9%). Adults 65 and older averaged 1.4 ± 3.3 needs; their most frequent needs were reported for the concept Sleeping (15.5%), Nutrition 13.1%), Digesting (9.5%), Home (9.5%), and Caretaking (8.3%), and Safe at home and work (8.3%) (Table 1).

Visually inspecting the images described in the methods section (Figures 1–4), we discovered six patterns and formulated and tested six hypotheses. All six hypotheses were supported (Table 2).

Figure 1.

Number of Strengths, Challenges, and Needs by Domain for Participant Adults 65+ and Adults 18-64

Figure 4.

Alluvial Flow Diagram of Respondents’ Number of Strengths, Challenges, and Needs Across Domains

Notes: 0 = No Strengths, Challenges, or Needs in any Domain; 1 = One Domain;

2 = Two Domains; 3 = Three Domains; 4 = Four Domains. Thin lines represent one respondent. Thick lines represent many respondents.

Table 2.

Pattern Description and Visualization Type

Pattern Description	Visualization Type	Hypothesis	Analysis (p value)
Pattern 1. The black bubbles (adults 65+) intersect or are within the grey bubbles (adults 18-64) (Figure 1).	This compound bubble chart depicts strengths, challenges, and needs by domain and age group.	No difference in strengths, challenges, and needs between adults 65+ and adults 18-64. This was supported.	Adults 65+ and adults 18-64 did not differ in the number of strengths, challenges, and needs by domain. (Chi square; p>.05 for all).
Pattern 2. The dashed line (strengths) is higher than the solid (challenges) and dotted (needs) lines across most of the x-axis (Figure 2).	This parallel coordinates graph depicts the percentages of strengths, challenges, and needs by concept.	Overall, adults 65+ would have more strengths than challenges and more challenges than needs. This was supported.	Adults 65+ had more strengths (mean number of strengths (19.8 ± 14.4) than challenges (7.4±7.9), and more challenges than needs (1.4 ± 3.3) (ANOVA, F=87.26, p<.001).
Pattern 3. The dashed line (strengths) slopes downward from left to right, and the solid (challenges) line slopes upward from left to right (Figure 2).	Parallel coordinates (variables same as above)	Concepts reported as strengths would be less likely to be reported as challenges. Likewise, concepts reported as challenges were less likely to be reported as strengths. This was supported.	Strengths and challenges by problem were negatively correlated (Pearson correlation, r=−.50; p<.001).
Pattern 4. The dashed line (strengths) slopes downward from left to right, and the dotted line (needs) is relatively flat (Figure 2).	Parallel coordinates (variable same as above)	Concepts reported as strengths would not be related to concepts reported as needs. This was supported.	Strengths and needs were not correlated for Adults 65+ (r= −.014, p=.899).
Pattern 4a. In the third box the mean is line higher and the box and whiskers are much larger than in the other boxes (Figure 3).	This box plot depicts the number of needs by group. Groups were formed by discretizing individuals into groups based on their number of strengths (0-10, 11-20, 21-30, and 31-42).	The group with the highest mean and largest box would be more likely to report a need. This was supported.	Adults 65+ with 21-30 strengths were more likely than the others to report needs (ANOVA, F=2.92, p=.04).
Pattern 5. The lines within all five segments on the two y-axes connected diversely with other segments on the left axis (strengths) and the middle axis (challenges) (Figure 4).	This Alluvial (Sankey) diagram depicts all interrelationships of each individuals’ strengths, challenges, and needs totals by domain for the sample.	There was no relationship between the number of strengths and challenges by domain. This was supported.	There was no association between strengths and challenges overall for adults 65+ (χ2, (1, N=84) = 0.24 p=.63).
Pattern 6. The lines within the lower segment of the middle axis (challenges) connect with the lower segments of the right axis (needs) (Figure 4).	Alluvial (Sankey) diagram (variable same as above)	Adults 65+ with challenges in four domains were more likely to have needs. This was supported.	Adults 65+ who reported challenges were more likely to report a need (χ2, (1, N=84) = 15.91, p<.001).

There are three stacked bubble charts in Figure 1, one each for strengths, challenges, and needs. Each layer in this stacked bubble chart compares data for across the four domains. The y-axis shows the number of strengths, challenges, or needs. The position on the x-axis corresponds to a domain. The sizes of the bubbles indicates the number of respondents in the two age groups (adults 18-64 and 65+). Larger bubbles indicate more respondents selected the particular strengths, challenge, or need. We named the pattern in Figure 1 “Pattern 1,” and hypothesized that the number of challenges, strengths, and needs did not differ by age group, which was supported.

There are multiple variables depicted in Figure 2, a parallel coordinates graph (Patterns 2, 3, and 4), with each of the SOST concepts arranged on the x axis in order of the greatest number of strengths. The percentages of respondents having a concept classified as a strength are connected by a dashed line (the top line). The percentages of respondents a concept classified as a challenge are connected by a solid line. The percentages of respondents having a concept classified as a need are connected by a dotted line (the bottom line). We named the patterns in Figure 2 “Patterns 2, 3, and 4.” For Pattern 2, we hypothesized older adults would have more strengths than challenges and more challenges than needs, which was supported. For Pattern 3, we hypothesized that strengths and challenges would be negatively corelated. For Pattern 4, we hypothesized that strengths and needs would not be correlated. All three hypotheses were supported.

Figure 2.

Percent of Adults 65+ Reporting Overall Strengths, Challenges, and Needs by Concept

The box plot in Figure 3 depicts strengths compared to needs for the sample stratified into four groups by their number of strengths (Pattern 4a). The pattern shows those with 21-30 strengths reported more needs than those in the other three groups. We tested the hypothesis that older adults in this group were more likely to report needs, which was supported.

Figure 3.

Distribution of the Number of Needs for Adults 65+ Classified into Groups by Number of Strengths

In the alluvial flow diagram (Patterns 5 and 6), each thin line represents one person (Figure 4). The thicker lines indicate more than one respondent with the same pattern. The segments on the three y-axes represent the total number of domains in which a respondent has strengths (left), challenges (middle), and needs (right). For example, starting on the top left, segment 0 shows adults 65 and older with strengths in no domains, and these lines connect with challenges in two, three, and four domains. Respondents in the top segment of the challenges axis have challenges in no (0) domains, and these lines connect with the top segment in the needs axis. Pattern 5 was the lack of a relationship between the number of challenges and strengths by domain. Pattern 6 was the higher likelihood of needs for respondents who reported challenges in all four domains. Both hypothesized patterns were supported.

Discussion

In this EDA study, the authors used a multistep data visualization pattern detection and testing process to examine whole-person health, focusing on adults 65 and older (Tukey, 1977). We formulated and tested hypotheses based on the six patterns identified in the visualizations; all six hypotheses were supported. These patterns revealed that older and younger adults in our sample did not differ by the number of strengths, challenges, or needs, that older adults had more strengths than challenges, and challenges than needs, that strengths and challenges by concept were inversely correlated, and that strengths and needs by concept were not correlated. A surprising pattern was that those with a high number of strengths reported the highest number of needs. The last two patterns identified were no relationship between the number of strengths and challenges by domain and more needs among respondents with challenges in all four domains. The pattern related to an inverse correlation between strengths and challenges by concept was similar to patterns noted in previous studies (Austin et al., 2021; Monsen et al., 2014; Monsen, Peters, et al., 2015; Radhakrishnan et al., 2016). The exception was the unexpected relationship between strengths and needs, for the subgroup with 21–30 of 42 maximum total strengths. The alluvial flow diagram revealed the uniqueness of each respondent. The use of structured Omaha System data simplified and expedited the visualization of whole-person health, including strengths, challenges, and needs across the four domains of health. These patterns should be tested using other datasets to validate the findings and further examine whole-person health patterns for diverse populations and settings.

Our findings of no differences in strengths, challenges, and needs between adults 65 and older and adults aged 18–64, and that neither group had a higher proportion of strengths compared to challenges and needs affirm that active older adults, such as those who may attend a state fair, are likely to have many strengths relative to their challenges and needs, similar to healthy adults ages 18-64 (Monsen et al., 2014; Monsen, Peters, et al., 2015). The finding of an inverse relationship between strengths and challenges across problems is intuitive and aligns with previous literature showing similar patterns in older adults (Monsen et al., 2014; Monsen, Peters, et al., 2015). The finding of respondent diversity across strengths, challenges, needs, and domains seen in the alluvial flow diagram underscores that all data are needed to truly understand whole-person health for individuals. Such data may inform clinical conversations, and aid in individualizing interventions that may leverage strengths to address challenges. This finding further reinforces the importance of incorporating each person’s values, preferences, and goals in care planning (Holt et al., 2020; Ortiz, 2018). The finding that older adults who reported challenges in three to four domains were more likely to report a need suggests that older adults with numerous health challenges may be more likely to have healthcare needs (and to express those needs) than those with fewer challenges (Cahill et al., 2009). Further research is needed to examine these and other clinically relevant implications of whole-person health data for older adults from the perspectives of clinicians, older adults, and family members (Zgraggen et al., 2018).

The finding of a positive relationship between strengths and needs for a subgroup with 21–30 strengths was unexpected. Given that these adults identified many strengths and needs, we posit that they may be more interested in improving their health than other subgroups within this sample. In other words, this may be evidence of patient engagement for adults 65 and older with many, but not all, strengths (Broderick & Haque, 2015; Faiola & Holden, 2017; Yi et al., 2018). This finding has implications for clinical care and patient engagement research relative to whole-person health.

From an informatics perspective, this study demonstrates the value of using Omaha System data because it is a rigorous taxonomy and ontology that describes all of health comprehensively and holistically (Martin, 2005). The use of Omaha System data is advantageous for visualization and other data-driven approaches because of the inherent specified relationships across the three components of the Omaha System that classify and measure whole-person health and healthcare interventions for a single discrete, defined problem list (Austin, Mathiason, et al., 2021; Martin, 2005; Peterson, 2017). Other existing datasets, such as clinical trial data, can be mapped to the Omaha System to make these relationships visible (Monsen et al., 2018).

This study had several limitations. As evidenced by their data, the adults were fair-goers who tended to be healthy and had several strengths. Selection bias was expected in this setting. To mitigate these limitations during data collection, research assistants were available to answer questions and provide on-the-spot technical support. Due to these limitations and the small sample size, the results are not generalizable. Further, this unique dataset was obtained in 2017. Given the global population health challenges that have occurred since that year, however, this dataset enables researchers to examine a relatively healthy sample of older adults from the community, providing a foundation for further research that will allow us to compare pre-pandemic health with more recent data generated during and after the COVID-19 pandemic. As in all EDA studies, it is prudent to examine multiple datasets to determine whether patterns are unique within the data or persist across multiple datasets (Zgraggen et al., 2018).

Conclusions

Through the use of data visualization, we identified patterns in whole-person health for adults 65 and older that were consistent with the literature, as well as clinically relevant unexpected findings. There is the potential to use these methods to understand a whole-person approach to health that includes strengths, challenges, and needs; and to advance research regarding the application of what we have learned in clinical settings. Findings show the value of using consumer-generated Simplified Omaha System Terms in research. These methods are suitable for future EDA research using diverse datasets.