Purpose

This project applies Long Short-Term Memory (LSTM) neural networks to predict potential equipment failure in hospitals using historical fault logs. The goal is to enhance maintenance planning and reduce sudden downtimes in clinical environments, thereby saving costs.

Why It Matters

Unexpected equipment failure can delay critical treatments and increase patient risk. According to the WHO, over 50% of medical equipment in developing countries is not fully functional. This project adapts LSTM which has been successful in aviation and manufacturing to predict failures in healthcare, supporting proactive maintenance, reducing downtime, and saving lives.

Purpose

A lightweight tool that helps people with diabetes track and visualise blood sugar patterns using easy-to-understand visuals. This project later evolved into DB GlucoTrack, a broader digital health platform.

Why It Matters

Diabetes is a global epidemic, affecting 1 in 10 adults worldwide (IDF, 2021). Many struggle to manage their condition due to lack of clear, actionable feedback. This tool helps bridge that gap by transforming blood glucose data into visual insights that users can understand and act on, especially vital in low-resource or self-managed settings.

Purpose

Conducted as part of my master's research, this project used bioinformatics tools and statistical analysis to identify potential biomarkers for early-stage Liver Cancer (Hepatocellular Carcinoma (HCC)) using publicly available gene expression datasets.

I identified three candidate biomarkers, one of which is AFP, an already clinically recognized biomarker for HCC, validating the accuracy of my analytical approach.

Why It Matters

Liver cancer is the third leading cause of cancer-related death globally, with a 5-year survival rate of just 18% mainly due to late detection (WHO, 2023). Existing biomarkers lack sensitivity and specificity. This research helps address the critical need for validated biomarkers for early-stage Hepatocellular Carcinoma, increasing chances of early diagnosis and treatment.