ARDSFlag: an NLP/machine learning algorithm to visualize and detect high-probability ARDS admissions independent of provider recognition and billing codes.

BACKGROUND: Despite the significance and prevalence of acute respiratory distress syndrome (ARDS), its detection remains highly variable and inconsistent. In this work, we aim to develop an algorithm (ARDSFlag) to automate the diagnosis of ARDS based on the Berlin definition. We also aim to develop a visualization tool that helps clinicians efficiently assess ARDS criteria. METHODS: ARDSFlag applies machine learning (ML) and natural language processing (NLP) techniques to evaluate Berlin criteria by incorporating structured and unstructured data in an electronic health record (EHR) system. The study cohort includes 19,534 ICU admissions in the Medical Information Mart for Intensive Care III (MIMIC-III) database. The output is the ARDS diagnosis, onset time, and severity. RESULTS: ARDSFlag includes separate text classifiers trained using large training sets to find evidence of bilateral infiltrates in radiology reports (accuracy of 91.9%±0.5%) and heart failure/fluid overload in radiology reports (accuracy 86.1%±0.5%) and echocardiogram notes (accuracy 98.4%±0.3%). A test set of 300 cases, which was blindly and independently labeled for ARDS by two groups of clinicians, shows that ARDSFlag generates an overall accuracy of 89.0% (specificity = 91.7%, recall = 80.3%, and precision = 75.0%) in detecting ARDS cases. CONCLUSION: To our best knowledge, this is the first study to focus on developing a method to automate the detection of ARDS. Some studies have developed and used other methods to answer other research questions. Expectedly, ARDSFlag generates a significantly higher performance in all accuracy measures compared to those methods.

Prevalence, genetic and clinical characteristics in first-degree relatives of patients with familial cerebral cavernous malformations in China.

OBJECTIVE: This study aims to investigate the prevalence of familial cerebral cavernous malformations (FCCMs) in first-degree relatives (FDRs) using familial screening, to describe the distribution of initial symptoms, lesion count on cranial MRI and pathogenic gene in patients. METHODS: Patients with multiple CCMs who enrolled from the Treatments and Outcomes of Untreated Cerebral Cavernous Malformations in China database were considered as probands and FDRs were recruited. Cranial MRI was performed to screen the CCMs lesions, and whole-exome sequencing was performed to identify CCM mutations. MRI and genetic screening were combined to diagnose FCCM in FDRs, and the results were presented as prevalence and 95% CIs. The Kaplan-Meier (KM) method was used to calculate the cumulative incidence of FCCM. RESULTS: 33 (76.74%) of the 43 families (110 FDRs) were identified as FCCM (85 FDRs). Receiver operating characteristic analysis revealed three lesions on T2-weighted imaging (T2WI) were the strong indicator for distinguishing probands with FCCM (sensitivity, 87.10%; specificity, 87.50%). Of the 85 FDRs, 31 were diagnosed with FCCM, resulting in a prevalence of 36.5% (26.2%-46.7%). In families with FCCMs, the mutation rates for CCM1, CCM2 and CCM3 were 45.45%, 21.21% and 9.09%, respectively. Furthermore, 53.13% of patients were asymptomatic, 17.19% were intracranial haemorrhage and 9.38% were epilepsy. The mean age of symptom onset analysed by KM was 46.67 (40.56-52.78) years. CONCLUSION: Based on MRI and genetic analysis, the prevalence of CCMs in the FDRs of families with FCCMs in China was 36.5%. Genetic counselling and MRI screening are recommended for FDRs in patients with more than three CCM lesions on T2WI.