Dr. 'Ismat Mohd Sulaiman

Dr. 'Ismat binti Mohd Sulaiman specializes in health informatics, semantic interoperability, and clinical text mining.

Connected Data, Transforming Pharma: Lessons from MyHDW and AI Potential

The Malaysian Health Data Warehouse (MyHDW) connects fragmented facility data into a national health repository built for big data analytics for health system planning. MyHDW treats data as a foundational asset to move forward towards predictive analytics and clinical text mining using AI. This talk examines what that connection teaches us about building an AI-ready system, and the potential using pharmacy data to take pharmaceutical care forward. MyHDW was developed in 2016 and has since accumulated years of experience from managing data from various healthcare facilities to producing insightful information for various agencies and researchers. The aim is to provide a high-level data-driven overview of the healthcare systems across the public and private sectors. That objective requires standardised data stored and shared in the same language, a governing body to create policies and ensure safety, as well as a continuous data quality assessment. MyHDW is equipped with health data standards for diagnosis and procedures using ICD, for health system planning using a national health data dictionary, and for deeper text-based analytics using SNOMED CT and LOINC. Reflecting on the pharmaceutical landscape, the same principles apply – the need for a nationally accepted drug data standard and governance. Otherwise, pharmaceutical data cannot move safely between facilities, and precision pharmacotherapy has nothing to build on. This talk will also present several pharma data-to-AI potential value chains if we do have drug data standards and governance. First, clinical decision supports systems are strengthened with medication safety through AI-supported interaction checking and individualised dosing. Second, pharmacovigilance and drug-utilisation analytics apply algorithmic methods to detect safety signals earlier than manual review allows. Lastly, predictive and personalised therapy uses pharmacogenomic frameworks to translate gene-drug data into actionable dosing guidance. Each depends on the same precondition: connected, standardised, governed data. Data governance must exist prior to AI governance, because governance ensures validation, testing, and monitoring of AI models built on top of the data. The same stewardship that makes MyHDW trustworthy must extend to how AI performs in practice. This talk ends with how the fraternity can prepare to embark on a safe and sustainable AI journey.