Anyone who has tried to extract usable data from a hospital's electronic health record knows the problem: physician notes are messy, inconsistent, and often buried in narrative text never meant for algorithmic eyes. Century Health, a New York startup that built its business around solving exactly that problem, just raised $5 million in a seed round to expand what it's calling disease-specific registries for pharmaceutical companies.
The funding, led by Origin Ventures, came together in an oversubscribed round with participation from InnovateHealth Ventures, 25madison, Next Play Ventures, continuing investors 2048 Ventures and Alumni Ventures, and a handful of strategic angels—clinicians working in nephrology, neurology, and ophthalmology who presumably understand the pain points firsthand. What caught investors' attention, according to the company, was a single number: 97% accuracy when measured against clinical expert judgment.
That figure, if it holds up under scrutiny, matters more than it might seem. Industry observers note that in a field where precision directly impacts drug development decisions and regulatory submissions, being wrong even a small percentage of the time can torpedo years of research.
Century's technology—branded as the Century Health Abstraction & Retrieval Model, or CHARM—is designed to automate the extraction of clinical variables from unstructured medical records while maintaining what the company describes as "full traceability" back to source documents. Think of it as teaching software to read like a clinical data manager, only faster and, ideally, more consistently.
Proving It Works
The company hasn't kept its methods entirely proprietary. In January, Century's team published research on arXiv detailing their work on prompt stability for clinical abstraction tasks. (The paper, titled "Stability-Aware Prompt Optimization for Clinical Data Abstraction," offers a glimpse into the technical architecture.) A subsequent white paper in May, focused on validating CHARM for real-world evidence applications, added more technical depth.
Whether that's enough to satisfy pharma's notoriously cautious data teams remains to be seen. But Century claims to have signed partnerships with multiple "Top 5" pharmaceutical companies—names it won't disclose publicly, a common refrain in an industry built on NDAs and competitive paranoia.
According to the company, its data network has grown 60x over the past year. That's an impressive rate, though it's worth noting that early-stage growth metrics can be deceptive when you're starting from a small base.
The Founders' Pedigree
Vish Srivastava and Sanjay Hariharan, who co-founded Century in 2023, bring complementary backgrounds to the problem. Srivastava previously led product management at Evidation and spent time at BCG; Hariharan came from Epic, the EHR giant, and later served as principal data scientist at McKinsey's QuantumBlack unit.
The founders' combination—clinical workflow expertise on one side, enterprise data infrastructure on the other—shows up in how Century positions itself. It's not trying to be a consumer health app or a telehealth platform. The company describes its approach as building plumbing for pharma.
Recent partnerships sketch out the company's specialty-focused strategy. A March collaboration with Dallas Renal Group aims to create what Century calls "one of the most detailed real-world datasets" for rare glomerular diseases in community care settings. In April, the company partnered with VCU's Stravitz-Sanyal Institute on steatotic liver disease. Earlier deals—with Nira Medical on multiple sclerosis data covering more than 3,000 patients, and Nimbus Health on respiratory conditions—follow a similar pattern: find a specialty area, integrate directly with EHR systems, and refresh the data monthly.
Century packages everything in OMOP Common Data Model format, the industry standard for interoperability, and claims HIPAA compliance and SOC 2 certification. (In other words, the basics you'd expect before any pharma company would touch your platform.)
Riding the Real-World Evidence Wave
The timing isn't accidental. Biopharma has been moving toward real-world evidence for years, driven partly by regulators who increasingly accept EHR and claims data for drug approvals. FDA guidance on using real-world data for drugs and biologics was updated in mid-2024 and again at the end of 2025, creating clearer—though still evolving—pathways.
A recent IQVIA Institute report noted a concentration of AI-related R&D deals in drug discovery, while regulators continue drafting guidance on AI in development submissions. Century is positioning itself at the intersection of those trends: AI-powered data extraction feeding into regulatory-grade evidence packages.
It's not the company's first fundraise. Century raised a $2 million pre-seed back in March 2024, led by 2048 Ventures with participation from Everywhere Ventures, LifeX Ventures, and Alumni Ventures. That round included angel checks from Travis May of Datavant and Christine Lemke of Evidation—both veterans of the health data world.
The Road Ahead
With the fresh capital, Century is hiring what it describes as a founding data scientist, a founding business development lead, and a fractional chief medical officer—all based in New York. The team page lists roughly a dozen employees, though the company's LinkedIn profile shows a 2-to-10 employee range. (That kind of lag is common among fast-growing startups; headcount numbers are always slippery.)
The funding positions Century to expand its provider network and add more disease areas. Its datasets already span renal, neurology, ophthalmology, respiratory, and metabolic conditions—a breadth that suggests ambition, if not yet complete market coverage.
But here's the question Century will need to answer: Can specialty depth and validated accuracy carve out durable differentiation in a real-world evidence market already crowded with established players?
Aetion, Komodo Health, Flatiron, ConcertAI—each has built its own data moats and pharma relationships over years, sometimes decades. Century's bet seems to be that purpose-built disease registries, refreshed monthly and powered by newer AI models, can offer something those incumbents can't quite match. Whether that's enough to build a defensible business, or just a feature that larger platforms will eventually absorb, is the kind of question that tends to get answered only after the seed capital runs out.
For now, at least, investors seem willing to find out.
