MedCycle MatchAI
Designing an AI system that proactively matches surplus medical supplies to clinics.
My Role
Product Designer
Team
4 team members
Design · ML · Product
Timeline
3-day design challenge
Tech for Good Basecamp
Organization
Background and problem
During the Silicon Valley Tech for Good Basecamp, I partnered with MedCycle Network, a nonprofit that redistributes surplus medical supplies from hospitals to safety-net clinics. Although large quantities of medical supplies are discarded every year, many community clinics still struggle to access the resources they need. I explored how machine learning could help MedCycle predict short-term local demand and proactively match donated supplies with clinics that need them most.
The Problem
In conversations with MedCycle staff, I learned that the biggest barrier wasn’t inventory tracking—it was the time required to find the right supplies. Clinic staff must manually search inventory and coordinate requests through emails and phone calls, which slows down distribution and often leaves usable supplies sitting unused.
Key challenges
• Clinics lack time to search inventory
• Matching relies on manual coordination
• Donations are unpredictable
• Supplies may remain unused while clinics still need them
My project process
Defining the Opportunity
After multiple stakeholder discussions, I clarified the complex ML challenge and mapped the problem space to understand where technology could create the most impact. MedCycle already collects valuable historical data—including clinic needs lists, donation records, and delivery outcomes—which creates an opportunity to predict which clinics are most likely to need specific supplies.
-
Clinics don’t have time to search inventory
-
Donations are unpredictable → matching requires fast decisions (“lumpy supply”)
Available data signals
• Clinic needs lists
• Wishlist items
• Donation records
• Historical deliveries
Solution
Based on the analysis before, after several discuss with team, We decided to design a ML model called MatchAI, an AI-powered recommendation system integrated into the MedCycle dashboard. I first designed and planned the logical structure and user flow of the feature we want to do. Instead of searching inventory, clinics receive recommended supplies predicted to match their needs. Staff can quickly review each item and decide whether to accept it. Currently Clinical Supply manager waste time they could spend treating patients on finding supplies. Donations are unpredictable, so person has to waste too much time checking and coordinating. So as the only designer, I brainstormed several potential solutions with our team and Non-Profit leaders and we finalized the ML that can identify which clinic needs what, and when
ML can identify which clinic needs what, and when
So as the only designer, I hand-drawn some low-fidelity prototypes on paper to make it easier to discuss with the team
Prototypes
I designed MatchAI,I hand-drawn some low-fidelity prototypes on paper to make it easier to discuss with the team
Our solution is that a Clinical supply manager arrives at their dashboard and sees "Daily Recommendations". Our ML model provides recommended items which we show in the list. User can click "Accept" and check the items, change the quantity and delivery date and click confirm once satisfied. Else they can click pass and items are removed from view.
They can click and check the items, change the quantity and delivery date and
This is where AI increases matching accuracy — clinics see only the items they are predicted to need, with quantities suggested.
Let’s see how a clinic manager accepts supplies in just 3 clicks.
Roadmap & Budget
I developed a detailed roadmap and budget plan to guide the implementation of the solution beyond the concept stage. This included defining key phases from ML logic design to MVP development, pilot testing, and full rollout, along with estimated timelines and resource requirements. By outlining a realistic path to production, I helped align stakeholders on feasibility and next steps, ensuring the solution could move from prototype to real-world impact.
Roadmap & Budget
I presented the final solution to MedCycle stakeholders, clearly communicating the problem, design rationale, and AI-driven approach. I translated complex ML concepts into an intuitive product narrative, helping stakeholders quickly understand how the system works and the value it could bring. The presentation played a key role in aligning stakeholders and contributed to the project receiving the “Most Complete Solution” Award.
Outcome & Impact
I presented the final solution to MedCycle stakeholders, clearly communicating the problem, design rationale, and AI-driven approach. I translated complex ML concepts into an intuitive product narrative, helping stakeholders quickly understand how the system works and the value it could bring. The presentation played a key role in aligning stakeholders and contributed to the project receiving the “Most Complete Solution” Award.
The final concept was well received by MedCycle stakeholders, and key elements of our solution were selected to inform future system improvements. Our team was also awarded the “Most Complete Solution” Award, recognizing the clarity, feasibility, and end-to-end thinking of our approach.
