Extending PICO with Observation Normalization for Evidence Computing.

While the PICO framework is widely used by clinicians for clinical question formulation when querying the medical literature, it does not have the expressiveness to explicitly capture medical findings based on any standard. In addition, findings extracted from the literature are represented as free-text, which is not amenable to computation. This research extends the PICO framework with Observation elements, which capture the observed effect that an Intervention has on an Outcome, forming Intervention-Observation-Outcome triplets. In addition, we present a framework to normalize Observation elements with respect to their significance and the direction of the effect, as well as a rule-based approach to perform the normalization of these attributes. Our method achieves macro-averaged F1 scores of 0.82 and 0.73 for identifying the significance and direction attributes, respectively.

A neuro-symbolic method for understanding free-text medical evidence.

OBJECTIVE: We introduce Medical evidence Dependency (MD)-informed attention, a novel neuro-symbolic model for understanding free-text clinical trial publications with generalizability and interpretability. MATERIALS AND METHODS: We trained one head in the multi-head self-attention model to attend to the Medical evidence Ddependency (MD) and to pass linguistic and domain knowledge on to later layers (MD informed). This MD-informed attention model was integrated into BioBERT and tested on 2 public machine reading comprehension benchmarks for clinical trial publications: Evidence Inference 2.0 and PubMedQA. We also curated a small set of recently published articles reporting randomized controlled trials on COVID-19 (coronavirus disease 2019) following the Evidence Inference 2.0 guidelines to evaluate the model's robustness to unseen data. RESULTS: The integration of MD-informed attention head improves BioBERT substantially in both benchmark tasks-as large as an increase of +30% in the F1 score-and achieves the new state-of-the-art performance on the Evidence Inference 2.0. It achieves 84% and 82% in overall accuracy and F1 score, respectively, on the unseen COVID-19 data. CONCLUSIONS: MD-informed attention empowers neural reading comprehension models with interpretability and generalizability via reusable domain knowledge. Its compositionality can benefit any transformer-based architecture for machine reading comprehension of free-text medical evidence.

A sex paradox in clinical outcomes following complex percutaneous coronary intervention.

BACKGROUND: Although the number of complex percutaneous coronary intervention (CPCI) procedures is increasing, data regarding sex-related outcomes following CPCI are scarce. METHODS: We retrospectively analyzed data of patients enrolled in a single-center registry between 2009 and 2017. Patients were divided into two groups (CPCI and non-CPCI) stratified by sex. CPCI was defined as any PCI procedure with ≥1 of the following characteristics: ≥3 target vessels/lesions, ≥3 stents implanted, bifurcation with ≥2 stents, stent length > 60 mm, or chronic total occlusion. The primary outcome was major adverse cardiac events (MACE), a composite of all-cause death, myocardial infarction (MI), and target vessel revascularization, at oneon-year follow-up. RESULTS: Among 20,419 patients, 5004 (24.5%) underwent CPCI of whom 25.6% (n = 1281) women and 74.4% (n = 3723) men. Women presented with more comorbidities yet less complex coronary anatomy than men (syntax score: 19.5 ± 10.3 vs. 20.6 ± 10.7, p = 0.009). Moreover, women were more likely to fulfill a single rather than multiple CPCI criteria. At one year, a higher rate of MACE occurred in women (14.0% vs. 11.6%, p = 0.02). After multivariable adjustment for confounders, the risk of MACE at one year was similar among both sexes (HR:1.04, 95% CI [0.85-1.26], p = 0.71), without significant interaction between the complexity of the procedure and sex (p-interaction = 0.96). Nonetheless, the risk of MI was significantly higher in women than men undergoing CPCI (HR:1.63, 95% CI [1.12-2.38], p = 0.01). CONCLUSIONS: Despite presenting with less challenging lesions than men, women had a higher rate of MI at one year following CPCI, even after adjusting for potential confounders.