BEYOND TRADITIONAL TESTING

How Swept AI made Forma Health the Most Trusted Name in AI-Assisted Rare Disease Research

"Healthcare demands we account for the unexpected—the edge cases and behaviors we couldn’t anticipate. We knew we needed a radically different approach to validation, which brought us to Swept."
Headshot of CEO of Forma Health

German Scipioni

CEO, Forma Health

image of an office collaboration scene (for a mobility and transportation)

The Trust Problem

As Forma Health prepared to deploy their AI solution, they faced a fundamental challenge that all AI companies must confront: traditional testing methods fail to address the dynamic, non-deterministic nature of AI systems. Or, put more simply, how do you validate systems that behave inconsistently?

The stakes couldn’t be higher in these clinical trials—a single corrupted data point can derail years of research and delay critical treatments. German Scipioni, CEO of Forma Health, deeply understands this pressure, which led him to seek out SweptAI. The goal: achieve an unmatched level of trust in their AI system's decisions.

"Standard unit and integration testing shows if your AI performs as expected in a single run, but AI systems can give different outputs even with the exact same inputs," explains Scipioni.

Making the Invisible, Visible

Swept approached this challenge with a comprehensive, two-part strategy

Phase 1: Pre-deployment Interrogation

Rather than running a fixed set of tests, Swept’s adaptive system actively probes AIbehavior through interrogation with tailored synthetic data. This approach explores thefull spectrum of AI decision-making, revealing the hiddent critical behaviors that noone thinks to look for.

Through this deep interrogation, Swept uncovered two critical findings:

  1. An agent systematic bias in pain scale interpretation that varied based on caregiver type, threatening the integrity of symptom data
  2. A potential privacy vulnerability in how the agents handled patient identifiers

"We were able to leverage Swept’s analysis to rebuild core components of our system tobe fundamentally more reliable," explains Scipioni.

What Swept discovered, frankly shocked us," says Scipioni.

"Their interrogation system found agent biases that could have corrupted our clinical trial data. The kind of issues you absolutely must catch before deployment."

Phase 2: Continuous Supervision

Next came the need for continuous validation of Forma Health’s AI system, asthese eveolve in production, requiring constant vigilance. They employed Swept'smonitoring platform to analyze both individual interactions and macro-levelpatterns, allowing them to intervene before a slight drift spirals out of control.

A small shift in how the system interprets medical terminology might seeminsignificant in a single case, but across thousands of patients, it could skew researchoutcomes.

Through Swept’s continuous monitoring, Forma maintains complete confidence in their system’s reliability. “In healthcare, you can’t wait for problems to become obvious,” notes Scipioni. “Swept gives us the visibility we need to ensure our AI remains trustworthy.”

“Traditional monitoring only tells you when something breaks. Swept allows us to step in long before that, when we’re able to nip the problem in”

Reliability-Driven Development

"Healthcare institutions require absolute confidence in AI systems," explains Scipioni. "Every decision must be explainable, every data point trustworthy. Swept's ability to find the hidden patterns gives us that confidence—and the evidence to back it”.

Swept's validation provided Forma Health with documented proof of their AI system's reliability, serving as crucial evidence for healthcare institutions and clinical research organizations. The partnership delivered:

Rapid deployment through efficient evaluation

Verified HIPAA compliance

Documented safety protocols

Elimination of bias in clinical data

The combination of thorough pre-deployment validation and continuous supervision has transformed how Forma Health is advancing rare disease research.As AI continues to transform medical research,Forma Health leads the way in demonstrating how powerful AI can be when backed by rigorous validation.In teaming with Swept AI, Forma Health has built the confidence to push boundaries in medical science, bringing new hope to rare disease patients.

The Results

  • Exposed critical flaws in AI decision-making that would have compromised critical research data
  • Ensured ongoing trust through continuous AI behavior monitoring
  • Accelerated path to market with comprehensive validation and certification in just two weeks
  • Maintained full regulatory compliance in complex healthcare environment

About Forma Health

Forma Health’s AI platform is changing the game for rare disease clinical trials by providing researchers and physicians with real-time insights into patient patterns. Their groundbreaking approach accelerates the path from research to treatment, bringing new hope to patient populations that traditional research methods struggled to reach.

Ready to Make Your AI Enterprise-Ready?

Schedule Security Assessment

For Enterprises

Protect your organization from AI risks

Get Swept Certified

For AI Vendors

Accelerate your enterprise sales cycle