MedShare

Turn Hospital Pharmacy Waste Into Savings

AI-powered pharmacy analytics that transforms spreadsheets into actionable intelligence — helping hospitals save thousands monthly on expired medications.

The Problem

Hospital pharmacies are hemorrhaging money on expired medications.

Statistic	Impact
*$800 billion	wasted annually on expired medications in US hospitals
20-50%	of prepared anesthesia drugs discarded unused
72%	of all drug returns are expired inventory
4x increase	in drug shortages over the past 5 years

The root cause? Pharmacy directors manage millions in inventory using spreadsheets and gut instinct.

No visibility into what's expiring until it's too late
No demand forecasting — just over-ordering "to be safe"
No FIFO compliance tracking — newer lots get used first, older lots expire
No analytics layer between inventory systems and decisions

"I manage 2,000+ medications in Excel. I don't know what's expiring until it's too late." — Hospital Pharmacist

The Solution

MedShare transforms pharmacy spreadsheets into actionable intelligence.

Upload CSV → See What's At Risk → Get AI Recommendations → Save Money

Upload your existing inventory export (CSV/Excel) — takes 5 seconds
Instantly see expiring medications ranked by dollars at risk
AI analyzes your usage patterns and forecasts 30-day demand
Get specific recommendations: "Reduce Propofol order by 15%"
Real-time voice alerts for critical expirations
One hospital. One upload. $4,200/month saved.

Key Features

Dashboard and Analytics

Real-time inventory overview with 4 key stat cards
Usage trend charts with 8-week historical data visualization
Department breakdown showing medication usage by OR, ICU, ER
30-day demand forecasting with confidence intervals

Expiration Alert System

Smart prioritization by urgency (High/Medium/Low)
Dollar impact calculation showing exact value at risk
FIFO violation detection preventing avoidable waste
CSV export for compliance and pharmacy director review

Voice Alerts

Dynamic voice announcements using Web Speech API
Live transcript display with word-by-word animation
Medicine-specific alerts with quantities and expiration dates

AI Drug Scanner

Webcam-based label recognition with computer vision
Automatic OCR for drug names, NDC codes, lot numbers
Quick inventory addition from physical labels

Multi-Hospital Support

Hospital selector with 3 demo locations (Metro General, St. Mary's, County Medical)
Distance-based matching for potential transfers
Network view showing nearby facility inventory

Reports and Exports

5 report types: Inventory, Expiration, FIFO, Forecast, Insights
PDF generation with professional formatting
CSV exports for all data views

What Makes It Revolutionary

Compound Intelligence Pipeline (Wood Wide AI)

We chain three ML models to create insights no single model could produce:

API	Function	Output
Cluster	Segments medications into risk profiles	High Risk, Stable, Volatile groupings
Predict	Forecasts demand per cluster	30-day usage with confidence intervals
Anomaly	Detects patterns within clusters	FIFO violations, usage spikes

Each output feeds the next — true compound reasoning, not sequential calls.

The insight loop:

Clusters tell you WHERE to look
Predictions tell you HOW MUCH is at risk
Anomalies tell you WHY waste is happening
Together: one actionable decision with dollar impact

Challenges We Ran Into

During development, one of the major challenges we encountered was integrating MongoDB with Atlas. Despite correctly configuring IP whitelisting and attempting multiple troubleshooting approaches—including switching networks, using mobile hotspots, and revalidating access rules—we were consistently blocked from establishing a stable connection. We escalated the issue to MLH representatives, who also attempted to diagnose the problem but were ultimately unable to resolve it due to the underlying nature of the Atlas networking constraints.

This obstacle significantly hindered our progress, as MongoDB was a core architectural choice intended to support a dynamic NoSQL schema for scalable ingestion of CSV-based pharmaceutical medicine inventory data. The flexibility of MongoDB would have allowed us to seamlessly accommodate varying data types and evolving inventory attributes. However, the persistent integration issues consumed valuable development time and began to impact our ability to deliver other planned features.

Rather than allowing this setback to stall the project, we demonstrated resilience and adaptability by pivoting our database strategy. We transitioned to Supabase, leveraging its PostgreSQL foundation and developer-friendly tooling to quickly restore momentum. Supabase integrated smoothly with our existing tech stack, provided reliable authentication and database access, and allowed us to model our inventory data with strong typing and relational integrity while still maintaining flexibility through JSON fields where needed. This strategic shift enabled us to unblock development, refocus on feature implementation, and deliver a stable, scalable backend solution despite the earlier infrastructure challenges.