As the premier gathering for infectious disease specialists, microbiologists, and public health practitioners in the Philippines, PSMID 2025 provided an essential clinical audience for our project. Our research, titled "Predictive Risk Scoring for Multidrug-Resistant Tuberculosis in the Philippines: A Machine Learning Approach," was presented in the poster session, focusing on the intersection of clinical microbiology and data science to improve diagnostic workflows for Multidrug-Resistant Tuberculosis (MDR-TB).

The research presented focuses on the development of a clinical decision-support tool that utilizes machine learning to assign a risk score to TB patients at the point of care. Unlike genomic approaches, this model prioritizes socio-demographic factors, clinical history, and previous treatment outcomes to identify individuals with a high probability of developing resistance. By quantifying risk through a standardized scoring system, the project aims to optimize the utilization of culture-based testing, ensuring that high-risk patients are prioritized in resource-constrained laboratory settings. The poster highlighted the model's high sensitivity and its potential to reduce the diagnostic delay that often leads to increased morbidity in MDR-TB cases.

Inquiries from attendees dove more into what data were used, highlighting the combination of clinical and lifestyle data as part of the risk computations, and emphasized the necessity of "interpretable AI"—models that provide clinicians with the specific features driving a high-risk score. This aligns with the convention’s theme of "Precision Medicine in Infectious Diseases," where data-driven insights are used to tailor antimicrobial therapy. Discussions with members of the PSMID Research Committee also touched upon the validation of the risk score across different regions, particularly in areas with high MDR-TB prevalence such as the National Capital Region and Calabarzon.

The inclusion of this work in the PSMID 2025 convention marks a significant step in the project’s clinical validation phase. While previous presentations focused on the technical architecture of the AI, this engagement focused on clinical utility and physician adoption. For the funding agency, this signifies that the project is successfully bridging the gap between computational research and bedside application. The insights gained from the convention will be used to refine the usability of the predictive tool, ensuring it meets the practical needs of Filipino clinicians and contributes effectively to the national strategy of ending TB.

The inclusion of this work in the PSMID 2025 convention marks a significant step in the project’s clinical validation phase. While previous presentations focused on the technical architecture of the AI, this engagement focused on clinical utility and physician adoption. For the funding agency, this signifies that the project is successfully bridging the gap between computational research and bedside application. The insights gained from the convention will be used to refine the usability of the predictive tool, ensuring it meets the practical needs of Filipino clinicians and contributes effectively to the national strategy of ending TB.

This national conference brought together stakeholders from academia, government, and the private sector to advance "implementation-ready" AI solutions across four strategic pillars: Social Good and Inclusive Growth, Governance and Disaster Risk Reduction, Cybersecurity, and BPO Transformation. Our project was selected for presentation under the Social Good and Inclusive Growth track, specifically contributing to the national discourse on AI-powered healthcare innovation and public health resilience.

The presentation, titled "Predicting the Unseen: Using Data to Protect Filipinos from Drug-Resistant Tuberculosis," detailed the development and preliminary findings of a predictive analytics framework designed to combat Antimicrobial Resistance (AMR), specifically the Multidrug Resistance Tuberculosis (MDR-TB). The research addresses a critical gap in the Philippine healthcare system: the significant delay between patient diagnosis and the identification of drug-resistant strains. By leveraging machine learning models trained on localized epidemiological data, the project demonstrated the ability to forecast resistance patterns with high accuracy. This allows for the immediate initiation of effective treatment regimens, thereby reducing the "time-to-correct-treatment" window and mitigating the risk of community-level transmission.

The project’s objectives are directly aligned with the Philippine Priorities for AI Social Good Framework, a strategic roadmap emphasized throughout the conference. Furthermore, the presentation highlighted how such AI-driven tools support the Department of Health’s (DOH) mandates for infectious disease control, specifically within the context of the National TB Control Program. The discourse emphasized that predictive modeling is not merely a technical achievement but a necessary intervention for safeguarding vulnerable populations in high-density urban areas.  The talk also highlighted the need for collaborative work across multiple institutions (private and public) in order to achieve implementation of the project.  Data security and readiness were topics of discussions as well since it was highlighted how the readiness of such data available in the Philippines can affect the development of any localized solution.