Natural Language Processing of Radiology Reports to Detect Complications of Ischemic Stroke.

BACKGROUND: Abstraction of critical data from unstructured radiologic reports using natural language processing (NLP) is a powerful tool to automate the detection of important clinical features and enhance research efforts. We present a set of NLP approaches to identify critical findings in patients with acute ischemic stroke from radiology reports of computed tomography (CT) and magnetic resonance imaging (MRI). METHODS: We trained machine learning classifiers to identify categorical outcomes of edema, midline shift (MLS), hemorrhagic transformation, and parenchymal hematoma, as well as rule-based systems (RBS) to identify intraventricular hemorrhage (IVH) and continuous MLS measurements within CT/MRI reports. Using a derivation cohort of 2289 reports from 550 individuals with acute middle cerebral artery territory ischemic strokes, we externally validated our models on reports from a separate institution as well as from patients with ischemic strokes in any vascular territory. RESULTS: In all data sets, a deep neural network with pretrained biomedical word embeddings (BioClinicalBERT) achieved the highest discrimination performance for binary prediction of edema (area under precision recall curve [AUPRC] > 0.94), MLS (AUPRC > 0.98), hemorrhagic conversion (AUPRC > 0.89), and parenchymal hematoma (AUPRC > 0.76). BioClinicalBERT outperformed lasso regression (p < 0.001) for all outcomes except parenchymal hematoma (p = 0.755). Tailored RBS for IVH and continuous MLS outperformed BioClinicalBERT (p < 0.001) and linear regression, respectively (p < 0.001). CONCLUSIONS: Our study demonstrates robust performance and external validity of a core NLP tool kit for identifying both categorical and continuous outcomes of ischemic stroke from unstructured radiographic text data. Medically tailored NLP methods have multiple important big data applications, including scalable electronic phenotyping, augmentation of clinical risk prediction models, and facilitation of automatic alert systems in the hospital setting.

A case of homicide by lethal injection with lidocaine.

This report describes a homicide by overdosage with lidocaine. The decedent, a 32-year-old male hospitalized for a lengthy period with acute intermittent porphyria and chronic pancreatitis, suffered apparent asystole and seizure. Failed resuscitation preceded death. A forensic autopsy was conducted based upon suspicions of alleged patient mistreatment by one of the attending nurses. Toxicological analyses revealed the presence of lidocaine in blood, liver, kidney, brain, and heart at 22.2 mg/L and 43.6, 28.3, 23.1, and 13.1 mg/kg, respectively. Also present were diazepam, phenytoin, and promethazine. Both diazepam and phenytoin had been administered during resuscitation, but lidocaine had not. The cause of death was determined to be ventricular arrhythmia precipitated by lidocaine overdosage. The administered dose was calculated to have been approximately 1500 mg. The manner of death was determined to be homicide. The nurse was arrested and subsequently tried for murder by administering a lethal quantity of lidocaine.

Incidence of fetal drug exposure in Alabama: 2004-2011.

Postmortem toxicological results for neonates and stillborn fetuses were examined to determine the incidence of drug exposure in utero and the relationship to fetal mortality. From 15,600 fatalities with toxicological analyses, 39 subjects were identified; 22 had positive findings (56%). Among the general population, 10,500 had positive findings (67%). Among the subject population, 22 drugs/metabolites were identified; cocaine/benzoylecgonine was most frequent (9 of 22 cases, 41%) then methamphetamine/amphetamine (7 of 22 cases, 32%). Respective incidences among the general positive population were 12% and 6%. Fetal/neonatal concentrations overlap the general population and exceed the mother. Remaining substances, detected largely in fatalities unrelated to drug exposure, were of little toxicological significance. The variety, with exceptions, reflects the general population. That the incidence of cocaine and methamphetamine among fetuses/neonates is three- and fivefold greater than the general population is evidence that maternal abuse of these drugs poses greater risk to fetal viability than all other substances identified.