Nuclear Medicine

LLM-assisted EMR in Kenyan primary care was safe but did not reduce 14-day treatment failure

In a pragmatic cluster RCT of 9,691 patients, LLM assistance did not significantly change 14-day treatment failure (2.2% vs 2.0%; adjusted OR 0.77, 95% CI 0.55–1.08; P=0.13). No serious adverse events were judged related to the LLM and independent review found no safety signal. The trial was conducted in 16 low-resource primary care facilities using EMR; effect sizes suggest any clinical benefit is modest. These results support cautious implementation with monitoring rather than widespread expectation of near-term outcome improvements. 1

Families report high involvement and satisfaction after receiving precision-medicine recommendations

In the PRISM trial, parents and adolescent patients expected benefit and recommendations; 70% of parents received a recommendation but only half recalled it. Parents reported high involvement (median 93/100) and satisfaction (95/100) with treatment decisions following recommendations. Receipt of a recommendation was not associated with regret about trial participation (p>0.05). Findings apply to families of adolescents with poor-prognosis cancer in a precision-medicine trial setting. 2

Radiomics AI predicts VETC and higher early-recurrence risk in HCC but evidence quality is limited

Meta-analysis of 15 studies showed CEMRI-based AI with sensitivity 0.84, specificity 0.79, and AUC 0.87 for VETC prediction. AI-predicted VETC was associated with increased early recurrence risk (pooled HR 2.34). Most data came from retrospective cohorts (primarily China and Japan) with low-to-moderate certainty due to bias and heterogeneity. Use these models as adjunct prognostic information, not a substitute for histopathologic CD34 assessment. 3