N-Power Medicine Patent Portfolio Analysis

The money arrives after the paper trail

A $32M Series B from Merck Global Health Innovation Fund, Innovatus Capital Partners and the LabCorp Venture Fund. Strategic money, all three of them, from people who buy clinical data and run clinical trials for a living. The company was founded in 2021 and describes itself as drug development and clinical research.

The last public filing is dated 2024-07-16. That is 26 months before this round closed. Anything filed since sits inside the eighteen-month publication window and is not visible yet. So the record you can read is the record as it stood well before these three investors committed, which makes the reading below a floor rather than a ceiling.

Two families, zero grants, one filing year

The granted ribbon is empty. Both families are published and pending, both filed in 2024, both in the United States only. Nothing here is enforceable today. First grants should land around 2027, which is roughly three years of runway during which the position is an application position and nothing more.

The titles are the interesting part. Systems and methods for automatic identification of anomalous data. Structured and unstructured data-driven classification. Neither of those is a drug, a dose or an assay. They are data-quality claims: catching the bad record, sorting the messy field. That is the plumbing underneath a clinical trial, not the trial itself.

The record you can read is a floor, not a ceiling.

One name on both, and a clinician beside it

Four inventors are named across the two families. Nils Gustav Thomas Bengtsson appears on both. Joseph Nathaniel Paulson and Christer Svedman appear on one each. That is a very small authorship graph, and its centre is a single person present at every filing decision the company has made so far.

Small graphs are cheap to read and cheap to lose. When one inventor carries the whole record, the institutional memory of why a claim was drafted a particular way lives with that person. It also tells you the filings were not a committee exercise. Two families, one recurring hand, one filing window: this was a deliberate, narrow, contemporaneous act.

The neighbourhood belongs to banks and search engines

The claims cluster in G06F. Over the last five years the same CPC groups drew Tencent Shenzhen with 3,273 families, Industrial & Commercial Bank Of China with 1,987, Beijing Baidu Netcom Sci & with 1,953, China Mobile Communications with 1,582, China Construction Bank with 1,453 and Ccb Fintech with 1,372. On that scale, two families is a dot.

Anomaly detection and mixed structured and unstructured classification are exactly what large banks and large platforms have been filing for a decade. That does not disqualify a small entrant. It does mean count is worthless as a measure here. What matters is whether the claims are written tightly to the clinical-data context, or broadly enough to be read against a very crowded prior art field.

Two citations, and neither is from pharma

Toronto Dominion Bank cites the portfolio twice. Vigilant Ai cites it once. Three citing publications, two organisations, both working in financial data integrity rather than drug development. Citation is not endorsement, but examiners and drafters pull the references that sit closest to what they are building. These two were looking at anomaly detection, and they found N-Power Medicine.

That is a quietly useful signal for a company with no grants. The applications are being read by people outside the industry the company says it serves. It suggests the claims were drafted in the general language of data rather than the narrow language of clinical operations, which is where scope questions get decided.

Hiring says biology, filing says software

Look at the 100 open roles. Forty-four are bio or clinical. Twenty-nine are other, thirteen are software or AI, ten are go-to-market, four are ops and finance. The sample titles run to contracts and legal operations, pharma delivery operations, clinical operations. That is a company staffing to run trials at scale. The patent record describes a company that cleans data.

Those two facts can be reconciled two ways. Either the 2024 filings capture the technical core and the clinical hiring is the delivery arm built around it, or the computing claims are the first product line and a second one is now being built that has not reached the public record. Both readings are consistent with everything visible. Only the next publication window separates them.

Two families in computing, forty-four open roles in biology.