BioMedizone Student Interns Use AI to Tackle Real-World Biomedical Challenges

Artificial intelligence is rapidly changing the way researchers approach medicine, and this summer, 17 high school students at BioMedizone explored how AI and machine learning could be used to solve real-world problems in biomedical science.The BioMedizone AI/ML Summer Research Interns recently completed their summer research program, presenting their projects at the organization's AI in Biosciences Summit 2026 on August 16, 2026. At the summit, interns presented their work to a panel of BioMedizone advisory board members, including Lauran Schnitkey, who is pursuing a PhD in pharmacology at Vanderbilt University, and Bitsar Banda, a chemical pathologist and medical scientist. Over the course of the summer, the 17 interns developed projects applying artificial intelligence to areas ranging from emergency medicine and medical imaging to genomics and blood-cell analysis.

Using AI to Improve Emergency Triage

Subscribe

One of the projects, VISOR (Visual & Imaging Integrated Score for Emergency Response), explored how AI could help healthcare professionals identify patients who may require critical care.This was worked on by students Diya Patel and Eshan Pradhan. The system combines clinical information such as vital signs and laboratory results with chest X-ray images. Using these two types of information together, VISOR predicts whether a patient may require escalation of care, including ICU admission or mechanical ventilation.

The project also incorporates Grad-CAM visual heatmaps and feature-attribution techniques, allowing users to see which portions of an X-ray and which clinical features contributed to the model's prediction.Rather than replacing healthcare professionals, the system was designed as a decision-support tool that could help make emergency assessments faster and more consistent.

Access the app: https://visor-surge-triage-767467039594.us-central1.run.app/

Exploring Disease-Causing DNA

Another project, Motifix, worked on Alimukhammed Samatuly, Sarvesh Shanthibooshan, Snehi Patel, Isha Uppala, Satvik Surapaneni, and Nour Aboul examined how deep learning could be used to identify potentially harmful patterns within DNA. The system uses a two-layer Bidirectional LSTM (Bi-LSTM) model and gradient-based saliency mapping to identify DNA motifs that may contribute to disease.

The project also looks at potential synonymous codon substitutions that could disrupt high-risk motifs without changing the amino acid sequence produced by the DNA. An interactive genomic explorer allows users to view sequences through annotated heatmaps and examine potential genetic changes through an accessible interface.

Bringing AI Into Pathology

Medical imaging was another major focus of this year's projects worked on by Navyaa Varma, Arina Potapenkom, Nysa Raina, Alaina Goyal and Veronica Benito. The Histological Morphological Mapper, or Slide Scope, uses deep-learning models to classify tissue from whole-slide histology images. The project fine-tuned ResNet-50 and Inception V3 models to distinguish between seven tissue categories, including mucus, muscle, normal mucosa and lymphocytes. The model achieved 99.5% validation accuracy and incorporated Grad-CAM heatmaps to show which cellular structures influenced its classifications. By highlighting relevant features while filtering background staining, the project demonstrates one possible approach to making AI-assisted pathology more interpretable.

Explore the model here: https://resnet-50-histologyzip--ariadna0.replit.app/

Detecting Blood Cells With AI

Meanwhile, SmearDx by Basit Arshid and Gauri, focused on microscopic blood-smear analysis. The project uses the lightweight YOLO26n object-detection model to locate, classify and count red blood cells, white blood cells and platelets in blood-smear images. After auditing annotations, cleaning the dataset and retraining the model, the system reached 85.4% mAP@50 and 82.46% precision. The researchers also implemented different confidence thresholds for each cell type to help the model identify rarer cells while limiting false detections.

Explore here: https://smeardx.vercel.app/

The Goal

While the projects covered very different areas of biomedical science, they shared a common goal: using computational tools to explore problems that researchers and healthcare professionals face in the real world. The summer program gave students the opportunity to move beyond simply learning about AI and instead apply machine-learning techniques to biomedical questions of their own. "Our 17 interns did a phenomenal job this summer, and we are excited to see how they transition into more long-term roles within BioMedizone after this internship," Prithi Balaji, Executive Director of Programs & Strategy at BioMedizone, said following the summit. The AI in Biosciences Summit marked the conclusion of the interns' summer research experience, but for many of the students, it is only the beginning of their work in biomedical research. BioMedizone hopes to continue giving students opportunities to explore the intersection of technology, medicine and scientific research while encouraging the next generation of biomedical innovators.

Be the next young innovator with us:https://www.biomedizone.org