Multimodal Deep Learning · Medical Imaging AI · Space Biology ML
Scroll to explore
Final-year CSE undergraduate at Daffodil International University, supervised by Dr. Md. Ali Hossain, Director of the NanoBio Technology Center.
My thesis addresses a real crisis — Bangladesh has fewer than 500 cardiologists for 170 million people. I built a trimodal deep learning system for CVD diagnosis from smartphone photos of paper ECGs, validated cross-nationally.
First AI/ML analysis of NASA OSD-970 — revealing 12.21-fold Ucp1 upregulation in female astronaut tissue after 37 days aboard the ISS.
Trimodal late fusion — ConvNeXt-Small CNN + XGBoost-ECG + XGBoost-Clinical — for five-class multi-label CVD classification. Original dataset of 1,191 patients at Enam Medical College, Dhaka. First multimodal smartphone-ECG dataset from Bangladesh. Cross-nationally validated without retraining (Pakistan CPEIC + NHFB Bangladesh).
First published ML re-analysis of NASA OSDR dataset OSD-970 (RR-1 ISS mission). RT-qPCR data, 89 genes, 16 female C57BL/6J mice, 37 days microgravity. Revealed 12.21-fold Ucp1 upregulation in WAT. Identified the Angpt2–Jun–Irs2–Klf2 axis via SHAP. Relevant to female astronaut health and obesity research.
Supporting ongoing laboratory research activities at the NanoBio Technology Center under Dr. Md. Ali Hossain. Co-organised the International Seminar on Smart Nanomaterials: Medicine, Energy, Material Science & AI.
Active volunteer for the SCARS initiative — organising winter clothing collection and distribution programs for underprivileged communities in Dhaka.
Volunteer for the "Smile More" community initiative, coordinating clothing and food distribution activities for disadvantaged children across Dhaka.
Seeking research internships, Masters scholarships, and academic collaborations in AI for healthcare and space biology.