2025
Global Health
-
<h6 style=”margin-top: -.5em;”>THE TEAM</h6>
<strong>Student Team:</strong> Mitch Lipke, Lindsay Lamberti, Santiago Sanchez, Jay Tailor, Selena Shirkin<strong>Advisors:</strong> Dr. Neha Verma, Dr. Shekhar Waikar Dr. Luis Soenksen, Dr. Samson Jarso, Adler Archer, David Bigio, Dr. Soumya Acharya, Dr. Youseph Yazdi, Bharat Shetty, Ganaraj Perunda, Sandhya Tiku
<h6 style=”margin-top: 2em;”>ABSTRACT</h6>
In rural India, clinical consultations are often constrained by time, resource limitations, and incomplete patient histories, leading to missed or delayed diagnoses. This project introduces an AI-driven, dual-agent simulation system designed to emulate realistic clinical encounters for the purpose of improving history-taking in low-resource telemedicine settings. The system features two independently controlled large language model (LLM) agents: a History-Taking Agent and a Simulated Patient Agent, which interact turn-by-turn within a structured controller framework. The HxTaker agent asks context-relevant questions and terminates when diagnostic sufficiency is reached; the patient agent responds based on a structured clinical case, avoiding unsolicited information. This agentic design mimics real-world interaction patterns, especially for patients with low literacy or complex symptom presentations.The system is optimized for Intelehealth’s mobile-based telehealth platform, supporting local languages, multiple-choice responses, and sub-10-minute session limits. Evaluation includes LLM judge scoring, AIME differential diagnosis accuracy, and structured output quality. Initial results show promising improvements in diagnostic yield (from 41% to 68%), with ongoing tuning of prompts and termination criteria. Future work includes integration of vital signs, trust scaffolds, and deployment in live clinical pilots to assess real-world impact and scalability in underserved healthcare settings.
-
THE TEAM
Student Team: Arjun Kumar Menta, Shreya, Jindal, Sofia Garcia del Barrio, Kenny Nova, Ava Taylor
Advisors: Dr. Soumya Acharya, Dr. Musoke Shariff, Dr. Sandra Sanchez, Dr. Mitala Yekosani, Dr. Sam Kalungi, Marina Rincon Torroella
ABSTRACT
Breast cancer is the leading cause of death for women worldwide, claiming 685,000 women’s lives annually. In low- and middle-income countries, the impact is particularly devastating, with survival rates plummeting to 40-60%, and 89% of diagnoses being late-stage. In Uganda, there exists a 6–11-month diagnostic delay that greatly exceeds the 60-days recommendation by the World Health Organization and significantly drives the high breast cancer mortality rate. The shortage and centralization of pathologists (18 pathologists:46M Ugandans), limited diagnostic ability in rural healthcare centers, cultural stigma, and the lack of financial means are the primary contributors to this crisis.
Ekyaalo Diagnostics aims to enable faster diagnostic turnaround times for breast cancer patients in low-resource settings by: (1) simplifying slide digitization, region-of-interest capture, and transmission for breast cytology; (2) streamlining communication with centralized, highly skilled hospitals; and (3) integrating our solution with the resources already available in these rural settings. Our low-cost technology leverages smartphones, existing microscopes, and eyepiece cameras to capture and transmit cytology images. These are sent via a mobile app powered by a robust machine-learning algorithm that highlights the most diagnostically relevant regions, enabling pathologists to work more efficiently and accurately.
-
THE TEAM
Student Team: Mark Schuweiler, Emma Turner, Sabrina Zhou, Mads Bundgaard Norlov, Amanda Cheund, Ishir Sharma
Advisors: Dr. Youseph Yazdi, Dr. Soumyadipta Acharya, Neetisha Bersa, Meghashish Sharma, Dr. Amita Gupta, Dr. Samyra Cox
ABSTRACT
Tuberculosis (TB) remains one of the leading infectious causes of death worldwide, with India bearing the highest burden. While national efforts have improved detection and treatment of active TB, identification of latent TB infection (TBI), especially in children, remains under addressed. Among India’s 250 million school-aged children, over half of pediatric TBI cases are estimated to go undetected yearly. This is largely due to the lack of scalable, field-deployable screening tools and limitations of the current standard: manual, clinician-dependent measurement of induration after TB skin testing (e.g., Mantoux or Cy-TB). This process is subjective, error-prone, and impractical in schools or homes without trained personnel. InduRate is a low-cost, user-friendly solution designed to transform pediatric TBI screening through objective, accessible measurement of TB skin test results. Any user—regardless of training—can press a deformable material against the forearm 48–72 hours post-test to capture an impression of the induration. Paired with a smartphone app, the system automatically determines if the test is positive or negative, removing user variability and clinician dependency.
Early validation with clinicians and untrained users has shown strong accuracy and usability. By enabling screening in schools and homes, InduRate addresses key barriers: subjectivity in measurement and limited access to care. It’s now being integrated into a school-based protocol with the JHU Gupta-Klinsky India Institute (GKII) and is designed to pair with antigen-based tests like Cy-TB, with the goal of increasing detection and preventing progression to active TB.
-
THE TEAM
Student Team: Antoine Noreau, Isabella Menendez, Natasha Mody, Saardhak Bhrugubanda, Sabine Meurs, Sneha Raj
Advisors: Dr. Soumyadipta Acharya, Sunny Patel, Marina Rincon Torroella, Dr. Neil Lobo, Mercy Opiyo, Dr. Michael MacDonald, James Kaweesa, Jovan Batte, Mike Kato, Bumali Kawesa, Kigongo Siriman, David Onanyang, Sean Blaufuss, Dr. Douglas Norris, Nathaniel Moller
ABSTRACT
VectorCam, an AI-enabled tool for high-throughput vector surveillance, was originally developed by a previous CBID team and validated through a randomized controlled trial (RCT). Building on their work, our team was tasked with two key objectives: (1) identifying strategies for scaling VectorCam to additional countries and new use cases, and (2) enabling effective utilization of the high-quality surveillance data it collects at national, district, and local levels.
To address the first objective, we conducted over 50 stakeholder interviews and a global landscape assessment to identify scaling opportunities, challenges, and strategic pathways. Using insights from this landscape analysis and interviews with expert entomologists, we developed VectorPAL, a unified decision support platform that integrates entomological, epidemiological, and climate/geographical data into actionable insights. VectorPAL enables health authorities to monitor vector dynamics and disease transmission risks in real-time at localized levels, facilitating targeted, timely, and sustainable vector control strategies. By bridging the gap between surveillance data and intervention planning, our work advances the ability of health systems to respond proactively to vector-borne disease threats worldwide.
Advanced Health
-
THE TEAM
Student Team: Mitch Lipke, Lindsay Lamberti, Santiago Sanchez, Jay Tailor, Selena Shirkin
Clinical Mentors: Dr. John McClure, Dr. Joshua Beckman, Dr. Clifford Weiss, Dr. Mark Lessne, Dr. Ali Tavallaei
Advisors:Mark Gelfand, Sean White, Kym McNicholas, Dr. Youseph Yazdi, Oriol Cuxart, Dr. Soumyadipta Acharya, Dr. Anders Sideris, Dr. April Zambelli-Weiner
ABSTRACT
Each year, over 150,000 Americans undergo major lower-limb amputation due to no-option Chronic Limb-Threatening Ischemia (CLTI), a condition for which standard revascularization is no longer viable. These patients face a 1-year mortality rate exceeding 25% and impose over $43 billion in annual U.S. healthcare costs. Current treatments—angioplasty, bypass, and venous arterialization—are ineffective in patients with heavily calcified, below-the-knee disease and rely on vessel patency and specialist access
BlueHealer is a novel, minimally invasive orthopedic device designed to restore perfusion from the bone marrow outward using distraction osteogenesis (DO). The system stimulates angiogenic signals, enhancing capillary formation and tissue healing. BlueHealer builds on proven Tibial Transverse Transport (TTT) methods but modifies the mechanism to reduce infection risk and patient burden. Early clinical interest, customer discovery, and a validated Class II De Novo regulatory path support its adoption. With a projected device cost of $20,000 and manufacturing costs under $4,000, BlueHealer offers a scalable, cost-effective alternative to amputation. Our next steps include prototype refinement, benchtop testing, cadaver studies, and FDA engagement, enabling future clinical trials and commercialization.
-
THE TEAM
Student Team: Arjun Kumar Menta, Shreya Jindal, Sofia Garcia del Barrio, Kenny Nova, Ava Taylor
Clinical Mentors: Dr. William Anderson, Dr. Risheng Xu
Advisors: Dr. Youseph Yazdi, Dr. April Zambelli-Weiner, Dr. Ashish Nimgaonkar
ABSTRACT
Traumatic brain injury (TBI) remains a leading cause of death and disability in the US, with an estimated 1.7M cases and 70K deaths annually. It mostly affects young people in vehicle accidents, and elderly in domestic falls. For the most severe cases, the only life saving intervention is a two-surgery treatment. During the first surgery, around half of the patient’s skull is removed for approximately 30-90 days to allow for unrestricted brain swelling. The second surgery replaces the skull portion that was removed, to provide the patient with long term brain protection.
These two surgeries are highly invasive, resulting in high costs (among the top 20 most expensive procedures in the US), significant complications, and lengthy hospital stays. Even after the swelling has subsided, patients must wait to have a second invasive surgery because they remain in such a fragile state from the first one, further delaying patient rehabilitation.
Neuro Prima’s novel cranial implant enables the functions of both surgeries within a single procedure. By eliminating the secondary surgery and associated infection rates, our solution enables a safer, faster, and more affordable recovery for patients.
-
THE TEAM
Student Team: Mark Schuweiler, Emma Turner, Sabrina Zhou, Mads Bundgaard Norlov, Amanda Cheund, Ishir Sharma
Clinical Mentors: Dr. Leila Mady, Dr. Anders Sideris, Dr. Dominick R. Guerrero
Advisors: Dr. Youseph Yazdi, Dr. Soumyadipta Acharya
ABSTRACT
SurgMap is developing a novel suite of tools to improve surgical outcomes for oral cavity squamous cell carcinoma (OCSCC), the most common form of head and neck cancer (HNC). OCSCC has a high incidence of about 35,000 cases annually, with a projected 40% increase by 2040. Patients are usually treated with a primary resection where excising adequate surgical margins around the tumor is critical for good outcomes. Clinical evidence shows that a close or positive margin left behind contributes to cancer recurrence. Unfortunately, cancer-free margins are achieved in only 15% to 26% of OCSCC cases. Thus, most patients with early-stage OCSCC will experience recurrence, requiring a second surgery or adjuvant therapy (e.g. radiation or chemotherapy).
Currently, surgeons rely on imprecise verbal descriptions and rudimentary tools to localize leftover tumor margins intraoperatively, often leading to inadequate removal or unnecessary damage to healthy tissue. Our team is developing a solution to map the anatomical position of a positive margin back to the resection bed, to reduce recurrence, prevent adjuvant therapy, and avoid unnecessary morbidity.
Our system includes SurgScan, a software platform that uses 3D scans of the resected tumor and tumor bed to allow pathologists to annotate margin locations, which are then spatially mapped back to the surgical field. Complementing this is SurgMark, a set of physical markers placed in both the specimen and tumor bed during surgery, enabling precise calibration and alignment between scans. Together, these tools enhance communication between surgeons and pathologists, improve resection accuracy, and help reduce both recurrence and functional impairments. SurgMap integrates into current surgical and pathological workflows, improving outcomes while lowering long-term healthcare costs.
-
THE TEAM
Student Team: Antoine Noreau, Isabella Menendez, Natasha Mody, Saardhak Bhrugubanda, Sabine Meurs, Sneha Raj
Clinical Mentors: Dr. Lukas Ramcharran, Dr. Michael Phelan, Dr. Michael Redlener
Advisors: Dr. Youseph Yazdi, Dr. Soumyadipta Acharya, Dr. April Zambelli-Weiner, Brisa Seibert
ABSTRACT
Hemolysis, the breakdown of red blood cells during sample collection, impacts approximately 17% of the 70 million blood draws conducted annually in emergency departments (EDs) across the United States. Current blood collection methods using IV catheters, syringes, and traditional vacuum tubes expose cells to excessive shear stresses, resulting in hemolysis that alters lab results, delays diagnoses by over an hour per patient, and creates an economic burden of approximately $15 billion annually due to increased hospital costs and prolonged patient wait times.
Veina Vascular introduces a novel blood collection device designed to optimize pressure differentials and to maintain laminar blood flow, dramatically reducing shear stress and subsequent hemolysis. Our device seamlessly integrates with existing IV catheters, requires minimal training, and ensures accurate lab results from a single draw. By preventing repeated blood draws, our technology improves clinical decision-making, shortens ED stays, and reduces hospital operational costs.
Developed through extensive clinician collaboration and validated by stakeholder interviews across North America and Europe, Veina’s solution directly enhances patient throughput, significantly improving emergency department efficiency and patient care quality.
