Our Machine Learning Engineer, Lin Lin, recently presented her industrial PhD thesis, addressing one of the most pressing challenges in global health: the persistent burden of infectious diseases, particularly in low- and middle-income countries (LMICs). Despite advancements in medicine and technology, diseases such as malaria, tuberculosis, HIV/AIDS, and neglected tropical diseases (NTDs)—including schistosomiasis, filariasis, and soil-transmitted helminths (STH)—continue to disproportionately affect vulnerable populations, perpetuating cycles of poverty and poor health outcomes.
Her research focuses on developing artificial intelligence (AI)-driven solutions to enhance the diagnosis of NTDs, particularly helminthiasis and filariasis. The project aims to address critical healthcare gaps in resource-limited settings, where shortages of trained medical personnel and technological constraints delay timely diagnosis and treatment.
This remarkable project results from a collaboration between SpotLab and the Universidad Politécnica de Madrid, in partnership with the Instituto de Salud Carlos III and the Kenya Medical Research Institute (KEMRI). The research received support from the Comunidad de Madrid and Spain’s Ministry of Science and Innovation.
Key Contributions:
- Development of unique datasets:
- A dataset for soil-transmitted helminthiasis with over 1,300 samples and approximately 20,000 images.
- A dataset for filariasis with 130 samples and 2,800 images.
- Innovative approaches to data labeling:
- Utilized crowdsourcing for image annotation, demonstrating that data labeled by non-experts can effectively train deep learning models.
- AI-powered diagnostic tools:
- Developed object detection algorithms for identifying helminthiasis eggs and filariasis, enabling seamless integration of the AI model into smartphones for portable diagnostics.
- Field validation:
- Conducted a field study during a deworming campaign in Kenya to assess the feasibility of deploying AI-driven tools in LMICs, addressing challenges such as infrastructure limitations, data availability, and user training.
- Advancements in filariasis diagnosis:
- Lab setting Validation: The filariasis detection algorithm underwent rigorous lab-setting validation, emphasizing human-AI interaction to maximize performance.
- Foundational models for fecal sample analysis:
- Established a pipeline for generating foundational models capable of diagnosing multiple parasites, enabling the development of more robust models with limited labeled data.
This groundbreaking work highlights the transformative potential of AI in microscopy-based diagnostics, the feasibility of deploying AI solutions in resource-constrained environments, and the progress toward building trustworthy AI systems. The techniques and methodologies developed in this thesis have the potential to be applied to other diseases, contributing to the achievement of universal health coverage and improving health outcomes globally.At SpotLab, we remain committed to harnessing technology to close gaps in global healthcare and provide innovative solutions for underserved communities. We congratulate Lin Lin on this extraordinary achievement and invite you to follow our journey on LinkedIn as we continue to advance the future of health diagnostics.