There is a recurring frustration in regulated industries: the data is most valuable when combined, and combining it is exactly what the rules forbid. Hospitals, insurers, and research networks each hold pieces of a picture — outcomes, claims, genomics — that would teach far more together than apart, but patient-privacy law and contracts keep the datasets siloed. Secure multiparty computation (MPC) is the cryptographer's answer.
US10824738B2, “Privacy-preserving analysis system for secure multiparty computing,” granted to HealthVerity, Inc. on November 3, 2020, applies that idea to health data. Its CPC tags — including G06F 21/6245 for protecting personal data and H04L 9/0894 for key distribution — read like a privacy-engineering bill of materials.
The core promise of MPC, and the thing the claim operationalizes, is that several parties can compute a joint function of their combined inputs while each input stays private to its owner. No party — and crucially no central aggregator — ever sees another's raw records; only the agreed-upon result emerges. The patent describes a system that orchestrates this for analytic queries over distributed sensitive datasets.
What makes the health setting a good fit is that the questions are often aggregate by nature: how many patients across these institutions match a profile, what is the distribution of an outcome, does a signal replicate across cohorts. These are precisely the queries MPC can answer without de-siloing the underlying records, which is why health-data networks were early and serious adopters.
Standard caveats from this desk: granted patent, kind code B2, not a pending application; and a system/method claim, not evidence of a particular product's market success. HealthVerity operates in exactly this real-world-health-data space, which is the plausible commercial context, but the IP is the technique.
The broader point for the IP map is that privacy-preserving computation stopped being purely academic around this period and started getting filed by operating companies with specific verticals — health here, finance elsewhere. When the assignees shift from universities to data businesses, that is usually the signal a technology is crossing from research into product.