Letter to the Editor
IN a modest side room at Kapit Hospital, Sarawak, a quiet revolution is taking shape. It’s not the kind that dominates headlines or sparks global trends — but the kind that saves lives, one pixel at a time.
Here, away from the bustle of city hospitals, staff prepare blood slides the old-fashioned way: a drop of blood, a smear across glass, carefully packed for delivery to a lab hundreds of kilometres away in Kuching. But what happens next is where the future begins.
At Universiti Malaysia Sarawak (Unimas), these slides are digitised, processed, and analysed using a homegrown artificial intelligence (AI) tool developed by a team at Universiti Malaya (UM). The system, called MalariaCare+, is already showing remarkable promise in detecting human malaria with greater speed and accuracy than ever before.
“This isn’t artificial intelligence for the sake of it,” says project lead Associate Professor Ir Dr Khairunnisa Hasikin. “This is applied intelligence; built with and for the people who deal with malaria on the ground.”
A shape-shifting enemy
Plasmodium knowlesi, a zoonotic strain of malaria once confined to macaques, is now the leading cause of human malaria in Malaysia, particularly in Sabah and Sarawak. Its resemblance to other Plasmodium species under a microscope, coupled with a rapid 24-hour replication cycle, makes timely and accurate diagnosis both urgent and difficult.
In rural clinics, where trained parasitologists are few and diagnostic tools are basic, misidentification is common. While AI-based tools for malaria already exist, many are too generic, overly complex, or trained on species common in other regions, not in Malaysia.
That gap is what drove the team at UM to develop MalariaCare+: a mobile-compatible, AI-powered diagnostic assistant designed specifically for local realities.
Building the engine first
With support from the AI for Medicine Research Grant Scheme 2025, the team set out to tackle the foundation: getting the AI to “see” with precision.
Their main achievement so far is the validation of a graph-enhanced YOLO model — a deep learning engine that can detect malaria-infected red blood cells, even in densely overlapping regions. Most image analysis systems struggle here, especially with low-quality slides from remote regions.
“We trained the AI to see what even the sharpest eyes might miss,” says Dr Khairunnisa. “It’s not about replacing doctors and clinicians. It’s about giving them a second set of expert eyes when they need it most.”
This AI model has already been embedded into a working app for human malaria detection —currently being tested using real blood smears sourced from Sarawak, including those from Kapit Hospital, and processed at Unimas.
A collaboration across campuses and clinics
Kapit Hospital may lack an advanced diagnostic lab, but it plays a crucial role in the pipeline: collecting patient samples and forwarding them to Unimas for staining, digitisation, and analysis.
This partnership between UM, Unimas, and frontline clinical settings is what makes the project not only functional, but highly timely.
“We didn’t build this in a vacuum,” notes Dr Khairunnisa. “It was shaped by conversations with doctors, technicians, and field researchers who know the gaps firsthand.”
The next phase, which is still in development, is the creation of a stage-specific classifier for Plasmodium knowlesi. This deep learning AI model will eventually help clinicians understand not just whether a patient is infected, but also what stage the parasite is in; a critical step for timely, personalised treatment.
This classfier will be embedded into an upgraded version of MalariaCare+ alongside features like real-time inference, visual explainability, and longitudinal patient tracking. The team is also working on making the app more field-ready: able to connect with smartphone-linked microscopes, store patient history securely, and operate with limited internet access.
Discussions are also underway to include MalariaCare+ in training modules at both Unimnas and UM, allowing future doctors, researchers and health workers to understand not just the biology of malaria, but also the power of technology in public health.
“This is about digital equity,” says Dr Khairunnisa. “AI tools shouldn’t be reserved for high-income hospitals. They should work for anyone, anywhere — especially in the places that need them most.”
A Malaysian model for global problems
Most medical AI tools are still focused on diseases like cancer in high-income countries. MalariaCare+ offers something rare: an interpretable AI system designed for neglected tropical diseases, built in a low-resource setting, by people who understand both the science and the local realities.
This effort directly supports Malaysia’s malaria elimination roadmap. But its real potential lies in how it could inspire cross-border adoption and collaboration throughout Southeast Asia and beyond.
Back in Kapit Hospital, another slide is carefully prepared, labelled, and placed into a box for its long journey to the lab. It’s a small act — but now, that slide has a powerful partner waiting on the other end.
This is the personal opinion of the author(s) and does not necessarily represent the views of DayakDaily. Letters to the Editor may be lightly edited for clarity.
