Nine clinicians reported lower consultation-preparation workload after using an AI summary tool in 33 oncology consultations, according to an Oct. 17 study in npj Digital Medicine. They estimated an average 3.26 minutes saved per case; the abstract says 63.6% of summaries were complete.
The system, called Artificial Intelligence-based First Consultation Preparation (AI-FCP), turns information in clinical records into structured summaries for consultation preparation. The researchers say they co-created it with end users. The authors acknowledge clinicians from the Netherlands Cancer Institute's Department of Head and Neck Oncology.
Nine clinicians completed a workload baseline before using AI-FCP; the researchers then followed 33 consultations prepared with the tool. On the NASA Task Load Index, a questionnaire about perceived workload, the reported overall score after use was 28.9% lower than baseline. The study gives a Hodges-Lehmann estimate difference of -12.3, with a 95% confidence interval from -21.7 to -3.3. Clinicians reported average time savings of 3.26 minutes per case (standard deviation 2.65 minutes).
The abstract separately reports successful preparation in 84.9% of consultations and complete summaries in 63.6%. It does not define the criteria for either measure in the accessible text, so the two percentages should not be treated as interchangeable.
The authors call the results exploratory and say further research is needed to test whether they generalize. The study had one group, involved nine clinicians and 33 consultations, and the authors say variation in clinic scheduling and enrollment may have introduced survivorship bias. Its reported measures concern workflow and clinician workload; the study does not establish improved diagnosis, treatment or patient outcomes.
Three authors disclosed employment at Kaiko.ai. The article says AI-FCP was not changed during the study sessions and that non-commercially affiliated authors collected, curated and analyzed the primary outcome data. The journal identifies the published text as an early, citable accepted version that will be replaced by the final Version of Record after further edits.
Sources
npj Digital Medicine study; illustrative photo: National Cancer Institute image 2528 (Bill Branson, public domain).
.jpg){kind=link}