NHS wants a second opinion on Palantir's glowing self-assessments
Supplier told to clear benefit claims with health chiefs, though enforcing approval could prove tricky
NHS England's CEO, Sir Jim Mackey, intends to instruct Palantir to substantiate its claims about the benefits of its Federated Data Platform (FDP) before future assertions are made. While acknowledging that enforcing this request may prove challenging, Mackey is prepared to communicate his expectations via a written letter. Healthcare MP Ben Coleman had previously cautioned against directly crediting improvements to the platform, but noted Palantir UK's senior vice president, Tom Watson, was making such claims.
Despite these assertions, Mackey admits enforcement could be difficult due to contractual limitations. The contract, awarded in November 2023, is worth £330 million and aims to improve planning and efficiency. However, concerns have arisen regarding the evidence supporting these claims. In July, a campaign group questioned figures used in Parliament, and the Office for Statistics Regulation (OSR) began scrutinizing the evidence.
Initial reports indicated that 139 trusts were utilizing the platform, with 137 reporting benefits. Yet, a subsequent Freedom of Information request revealed that a third of trusts using Palantir's health data platform were experiencing fewer patient procedures than prior to implementation. The OSR subsequently emphasized the need for clearer communication of FDP performance claims.
Palantir's UK senior vice president, Tom Watson, defended the platform, citing benefits from various hospitals but omitting caveats regarding causality. Watson noted that while the OSR pointed out the absence of studies controlling for external factors, he emphasized ongoing benefits from hospitals. An independent evaluation of the FDP by Imperial College Projects is scheduled for February 2026, with results expected after the decision on contract extension, which expires in March 2027.
Written by urgent.news from The Register's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.