Synergizing Artificial Intelligence With Clinicians' Expertise: Towards Personalized Proactive Diabetes Care in Malaysian Primary Health Clinics
https://doi.org/10.15605/jafes.041.S1
- Author:
Sindeh Woweham
1
;
Mohd Nazri Mohd Daud
1
;
George George Mathew
1
;
Hazwani Hanum Hashim
1
;
Khairun’naim Khairuddin
1
Author Information
1. Faculty of Medicine and Health Sciences, Universiti Malaysia Sabah
- Publication Type:Journal Article
- MeSH:
Artificial Intelligence;
Diabetes Mellitus
- From:
Journal of the ASEAN Federation of Endocrine Societies
2026;41(S1):32-33
- CountryPhilippines
- Language:English
-
Abstract:
Introduction:Diabetes affects 15.6% of Malaysian adults and remains a
leading cause of complications and mortality. Although
locally developed artificial intelligence (AI) tools have
shown strong technical performance, their integration
with clinicians’ decision-making in primary care remains
fragmented. This perspective proposes a synergistic AI–
clinician framework to deliver truly personalized proactive
diabetes care.
Methodology:A multidisciplinary team designed a clinical workflow that
integrates three validated Malaysian AI components: (1)
LightGBM (LGBM) model trained on Malaysian National
Diabetes Registry (MNDR) data for complication risk
prediction, (2) DR.MATA AI retinopathy screening with
automated triage, and (3) AI-enhanced mobile apps for
real-time self-management. These layers feed into a single
secure clinician dashboard. The framework was modelled
using existing datasets (MNDR 2011–2021, NHMS 2023,
DR.MATA validation cohort >14,000 images) and aligned
with the National AI Roadmap 2021–2025 and the
forthcoming National AI Action Plan 2026–2030.
Results:The LightGBM model achieved ROC-AUC values of 0.84
(mortality), 0.71 (retinopathy/nephropathy), 0.66 (ischemic
heart disease), and 0.74 (stroke). DR.MATA demonstrated 93.3% accuracy with automated triage. When integrated,
modelling suggests the framework could reduce major
complications by approximately 20–30% and unnecessary
referrals by up to 35%. The system runs automatically upon
opening the patient record, requiring no extra clinician
steps.
Conclusion:Synergizing AI with clinicians’ expertise offers a scalable
solution for Malaysian primary health clinics. A 12-month
national pilot is recommended to generate real-world
evidence and inform the National AI Action Plan 2026–2030.
- Full text:2026072815272775228EP_A026.pdf
