Can AI bolster radiologists?

Our client, a large European private equity manager, is considering an investment into a leading UK telediagnosis platform focused on outsourced radiology reporting services to the national health service and independent healthcare providers through a network of over 300 freelance consultant radiologists.

Complexity / Current state

The Target was a leading UK-based telediagnostics platform seeking new capital to continue its growth ambitions. In a competitive process, our client – a large, European private equity manager – was seeking technology, data & AI due diligence to investigate the Target’s capability and capacity for continued growth through operational efficiencies and effectiveness.

Recommendation / Use case

Processing tens of thousands of scans a year, the Target was processing large volumes of data (ownership of which was retained by the national health service Trusts). Data latency, plus the need for enhanced machine learning operations were opportunities identified to improve in-demand, time-sensitive diagnosis.

Outcome

Our due diligence concluded that AI cannot be used to automate the work of radiologists (they are legally responsible), but it can be used to augment their work (making them more efficient).

The Target was acquired by an alternative, PE-backed buyer.

With record demand for imaging services, the Target has grown rapidly – thanks, in part, to a well invested tech platform and workflow process that is intended to embed scalability, efficiency, and competitive advantage.

With additional growth and client acquisition (plus, cross-sell and up-sell predicted), the private equity manager wished to test the extent to which the current platform and the wider tech-stack represents a suitably architected, resilient, secure, and scalable solution for today’s volumes. And, as a platform for further local and international growth, plus the adoption of greater AI capability.

Working alongside technology due diligence services providers, with whom DataDiligence has collaborated on multiple occasions, we conducted a data & AI vendor due diligence process to inform the investment decision, and provide insight for the investor into opportunities, risks, and investment required.

The Target was processing ~ one million scans every year, and this number was growing due to demand – particularly urgent reporting services (such as night-time, and time-sensitive reporting). While the Target had introduced an AI strategy (to support reporting, prioritisation, efficiency, productivity gains, greater accuracy, and better patient outcomes) plus upgraded its centralised systems to be AI ready – the actual integration of AI tools was time-consuming, particularly when seeking approval for new data processing.

As a regulated sector, it was pleasing to note that data ownership was retained by the national health trusts – thereby easing potential legislative burdens. However, this lack of data ownership could limit the Target’s ability to build, train, and test its own AI solutions. Also, as data latency was causing material time pressure, several infrastructure recommendations were suggested which would improve connectivity (particularly to external data sources) and processing speed – as is so important in trauma diagnosis, and other time-sensitive situations. Furthermore, designing machine learning operations around the AI tools would enable automated evaluation and feedback mechanisms to monitor the performance and functionality of each AI algorithm, again with the view to improving efficiencies.

Follow a comprehensive due diligence process, the private equity investor did not submit a bid. The Target was ultimately acquired by a different, private-equity backed trade buyer.