Electronic Health Records Reveal Three Key Patterns Linked to Diabetic Retinopathy

Ilustration By AI

FORMOSA NEWS - Purwokerto - Electronic health records could help healthcare providers identify diabetic patients who may need earlier screening for diabetic retinopathy, according to Mukhamad Hasim Iswanto of UIN Profesor Kiai Haji Saifuddin Zuhri Purwokerto. Published in 2026 in the Formosa Journal of Science and Technology, the study analyzed 30 electronic health records from diabetic patients treated at referral healthcare facilities in Indonesia during January–December 2025. The analysis found a recurring combination of longer diabetes duration, higher glucose levels, and high blood pressure among patients with indications of diabetic retinopathy. The findings highlight how routine medical records could support screening priorities without replacing eye examinations or specialist evaluation.

Diabetic Retinopathy Can Develop Before Vision Problems Appear

Diabetic retinopathy is a complication of diabetes caused by progressive damage to the small blood vessels of the retina. In its early stages, retinal damage may develop before patients notice significant changes in their vision. Delayed screening can therefore allow the condition to progress before it is detected.

The scale of the problem is substantial. The article cites a global systematic review estimating that approximately 103.12 million people with diabetes experienced diabetic retinopathy in 2020, with the number projected to reach about 160.50 million by 2045.

The situation is also relevant in Indonesia. Previous research cited by Iswanto reported an incidence of diabetic retinopathy of 34.6 cases per 1,000 person-years among Indonesian adults with type 2 diabetes, while vision-threatening diabetic retinopathy reached 24.5 cases per 1,000 person-years. Longer diabetes duration was also associated with increased risk.

These figures reinforce the importance of identifying patients who may require closer monitoring. Electronic health records offer a potentially useful source because they already contain information about diabetes history, glucose levels, blood pressure, demographics, comorbidities, medications, and other clinical indicators.

Turning Routine Medical Records Into Clinical Patterns

Mukhamad Hasim Iswanto used an exploratory descriptive approach based on secondary electronic health record data. The analysis focused on finding recurring relationships among patient characteristics, diabetes duration, glucose levels, blood pressure, and recorded retinopathy conditions.

The dataset consisted of 30 electronic health records selected through purposive sampling. Records were included when they contained relevant information on age, sex, duration of diabetes, glucose levels, blood pressure, and retinopathy status.

The researchers organized and cleaned the records before analyzing them. Microsoft Excel was used for initial tabulation, while Python supported the grouping and exploration of clinical patterns. The approach was designed to understand relationships among the available variables rather than develop a validated diagnostic or prediction model.
Among the 30 records, 18 patients were men and 12 were women. The average patient age was 57 years. Fourteen patients, or 46.7 percent, had indications of diabetic retinopathy, while 16 patients, or 53.3 percent, had no recorded indications of retinopathy.

Three Factors Stood Out

The strongest differences between the two groups involved diabetes duration, glucose levels, and blood pressure.

1. Longer diabetes duration

Patients with indications of diabetic retinopathy had an average diabetes duration of 13.1 years, compared with 6.7 years among patients without indications.

Among patients with retinopathy indications, 92.9 percent had lived with diabetes for at least 10 years. In comparison, only 12.5 percent of patients without retinopathy indications had diabetes for 10 years or longer.

2. Higher glucose levels

The average glucose level was 221.2 mg/dL among patients with retinopathy indications, compared with 161.1 mg/dL among those without indications.

A total of 85.7 percent of patients with retinopathy indications had glucose levels of at least 200 mg/dL. Only 12.5 percent of patients without retinopathy indications fell into the same category.

3. High blood pressure

High blood pressure was found in 85.7 percent of patients with retinopathy indications, compared with 25 percent in the group without indications.

Systolic blood pressure also differed substantially. The average was 151.9 mmHg in the retinopathy-indication group and 127.3 mmHg among patients without indications.

The Combination Was More Informative Than Individual Factors

The most prominent pattern emerged when the three factors were considered together.

Of the 14 patients with indications of diabetic retinopathy, 10 patients, or 71.4 percent, simultaneously had diabetes for at least 10 years, glucose levels of at least 200 mg/dL, and high blood pressure. No patient in the group without retinopathy indications showed all three factors simultaneously in the analyzed dataset.

The finding suggests that electronic health records may provide more useful information when clinical variables are interpreted collectively rather than separately.

In an ethical paraphrase of the study's interpretation, Mukhamad Hasim Iswanto of UIN Profesor Kiai Haji Saifuddin Zuhri Purwokerto explains that diabetes duration represents the chronicity of the disease, glucose levels reflect metabolic conditions, and blood pressure represents vascular burden. When these three dimensions appear together, they form a more meaningful clinical profile for identifying patients who may require earlier attention.

Potential Impact on Healthcare Screening

The practical value of the findings lies in the possibility of using information already stored in electronic health records to support screening priorities.

Healthcare facilities could potentially flag diabetic patients who have a combination of long disease duration, elevated glucose levels, and high blood pressure for further retinal assessment. Such an approach could help clinicians make better use of routine clinical information, particularly in settings where resources for screening are limited.

However, the researchers explicitly caution against treating the identified pattern as a diagnostic tool. The findings do not replace fundus examinations or ophthalmological evaluations. Instead, they could serve as an additional source of information for clinical decision support.

The study also has important limitations. Its analysis involved only 30 electronic health records selected purposively, meaning the findings cannot yet be generalized to the wider diabetic population. The dataset also lacked several potentially important variables, including longitudinal HbA1c measurements, lipid profiles, body mass index, kidney function, diabetes treatment, and retinopathy severity.

Because the research was exploratory and descriptive, it also cannot establish cause-and-effect relationships. The authors recommend larger, multicenter, longitudinal studies and external validation before these patterns are developed into clinical decision-support systems.

Author Profile

Mukhamad Hasim Iswanto
Affiliation: UIN Profesor Kiai Haji Saifuddin Zuhri Purwokerto
Field reflected in the article: data mining, electronic health records, health informatics, and early identification of diabetic retinopathy.

The journal metadata does not provide an academic degree for Mukhamad Hasim Iswanto, so no degree is added here to avoid introducing unsupported information.

Research Source

Article title: “Data Mining Approach for Early Identification of Diabetic Retinopathy Using Electronic Health Records”
Journal: Formosa Journal of Science and Technology
Publication year: 2026
Volume: 5, No. 8, pp. 2265–2280
Author: Mukhamad Hasim Iswanto, UIN Profesor Kiai Haji Saifuddin Zuhri Purwokerto

Posting Komentar

0 Komentar