PhD defence B. (Bart-Jan) Boverhof

Towards Valuable Artificial Intelligence in Healthcare and its Assessment

On Thursday 10 December 2026 B. Boverhof will defend the doctoral thesis titled: Towards Valuable Artificial Intelligence in Healthcare and its Assessment

Promotor
Prof.dr. M.P.M.H. Rutten - van Molken
Promotor
Prof.dr. C.A. Uyl - de Groot
Co-promotor
Dr. W.K. Redekop
Co-promotor
Dr. J.J. Visser
Date
Thursday 10 Dec 2026, 10:30 - 12:00
Type
PhD defence
Space
Senate Hall
Building
Erasmus Building
Location
Campus Woudestein
Add to calendar

Below is a brief summary of the dissertation:

Artificial intelligence (AI) can read scans and spot disease, sometimes as accurately as doctors. Yet most medical AI never gets past the pilot stage. A high accuracy score says little about whether patients get healthier, whether hospitals can work with it, or whether it is worth the money. This dissertation asks how we can tell which medical AI is truly valuable.

First, it offers two tools for judging that value. RADAR is a seven-step ladder for radiology AI, from "does the software work technically?" to "does it help patients, is it worth the cost, and does it work in this particular hospital?". A broader checklist of 29 points adds questions about workflow, staff, privacy, fairness, liability and transparency.

Second, it looks at the money. A review of 117 economic studies showed that most evaluated AI that was still immature, and that costs such as implementation and maintenance were often left out. A model of the Dutch population then showed that AI-based glaucoma screening for people aged 50 to 75 could be cost-effective, at about €19,000 per healthy life-year gained. AI makes this possible by acting as a cheap first filter, turning screening that used to be unaffordable into a realistic option.

Third, the dissertation shows that AI can also speed up the assessment process itself. An AI tool read 49 European assessment reports (over 5,000 pages) and pulled out the key questions for evaluating new cancer treatments. It performed as well as or better than human reviewers, saving about 25 hours of work for minimal costs.

The overall message: medical AI is neither automatically good nor inherently harmful. Its value depends on how it is developed, evaluated and governed.

More information

The public defence will begin exactly at 10.30 hrs. The doors will be closed once the public defence starts, latecomers may be able to watch on the screen outside. There is no possibility of entrance during the first part of the ceremony. Due to the solemn nature of the ceremony, we recommend that you do not take children under the age of 6 to the first part of the ceremony.

A live stream link has been provided to the candidate.

Compare 0 study programme