Deep learning neural network derivation and testing to distinguish acute poisonings.

INTRODUCTION: Acute poisoning is a significant global health burden, and the causative agent is often unclear. The primary aim of this pilot study was to develop a deep learning algorithm that predicts the most probable agent a poisoned patient was exposed to from a pre-specified list of drugs. RESEARCH DESIGN & METHODS: Data were queried from the National Poison Data System (NPDS) from 2014 through 2018 for eight single-agent poisonings (acetaminophen, diphenhydramine, aspirin, calcium channel blockers, sulfonylureas, benzodiazepines, bupropion, and lithium). Two Deep Neural Networks (PyTorch and Keras) designed for multi-class classification tasks were applied. RESULTS: There were 201,031 single-agent poisonings included in the analysis. For distinguishing among selected poisonings, PyTorch model had specificity of 97%, accuracy of 83%, precision of 83%, recall of 83%, and a F1-score of 82%. Keras had specificity of 98%, accuracy of 83%, precision of 84%, recall of 83%, and a F1-score of 83%. The best performance was achieved in the diagnosis of single-agent poisoning in diagnosing poisoning by lithium, sulfonylureas, diphenhydramine, calcium channel blockers, then acetaminophen, in PyTorch (F1-score = 99%, 94%, 85%, 83%, and 82%, respectively) and Keras (F1-score = 99%, 94%, 86%, 82%, and 82%, respectively). CONCLUSION: Deep neural networks can potentially help in distinguishing the causative agent of acute poisoning. This study used a small list of drugs, with polysubstance ingestions excluded.Reproducible source code and results can be obtained at https://github.com/ashiskb/npds-workspace.git.

Classification of acute poisoning exposures with machine learning models derived from the National Poison Data System.

The primary aim of this pilot study was to develop a machine learning algorithm to predict and distinguish eight poisoning agents based on clinical symptoms. Data were used from the National Poison Data System from 2014 to 2018, for patients 0-89 years old with single-agent exposure to eight drugs or drug classes (acetaminophen, aspirin, benzodiazepines, bupropion, calcium channel blockers, diphenhydramine, lithium and sulfonylureas). Four classifier prediction models were applied to the data: logistic regression, LightGBM, XGBoost, and CatBoost. There were 201 031 cases used to develop and test the algorithms. Among the four models, accuracy ranged 77%-80%, with precision and F1 scores of 76%-80% and recall of 77%-78%. Overall specificity was 92% for all models. Accuracy was highest for identifying sulfonylureas, acetaminophen, benzodiazepines and diphenhydramine poisoning. F1 scores were highest for correctly classifying sulfonylureas, acetaminophen and benzodiazepine poisonings. Recall was highest for sulfonylureas, acetaminophen, and benzodiazepines, and lowest for bupropion. Specificity was >99% for models of sulfonylureas, calcium channel blockers, lithium and aspirin. For single-agent poisoning cases among the eight possible exposures, machine learning models based on clinical signs and symptoms moderately predicted the causal agent. CatBoost and LightGBM classifier models had the highest performance of those tested.

Pediatric Fatalities Associated With Over-the-Counter Cough and Cold Medications.

BACKGROUND AND OBJECTIVES: In 2008, over-the-counter cough and cold medications (CCMs) underwent labeling changes in response to safety concerns, including fatalities, reported in children exposed to CCMs. The objective of this study is to describe fatalities associated with exposures to CCMs in children <12 years old that were detected by a safety surveillance system from 2008 to 2016. METHODS: Fatalities in children <12 years old that occurred between 2008 and 2016 associated with oral exposure to one or more CCMs were identified by the Pediatric Cough and Cold Safety Surveillance System. An expert panel reviewed all cases to determine the causal relationship between the exposure and death, if the intent of exposure was therapeutic, and if the dose was supratherapeutic. Other contributing factors related to the child's death were also identified as part of a root cause analysis. RESULTS: Of the 180 eligible fatalities captured during the study period, 40 were judged by the expert panel to be either related or potentially related to the CCM. Of these, the majority (n = 24; 60.0%) occurred in children <2 years old and involved nontherapeutic intent (n = 22; 55.0%). The most frequently involved index ingredient was diphenhydramine (n = 28; 70.0%). In 6 cases (n = 6; 15.0%), the CCM was administered to murder the child. In another 7 cases (n = 7; 17.5%), death followed the intentional use of the CCM to sedate the child. CONCLUSIONS: Pediatric fatalities associated with CCMs occurred primarily in young children after deliberate medication administration with nontherapeutic intent by a caregiver.

Impact of product safety changes on accidental exposures to liquid laundry packets in children.

OBJECTIVES: To evaluate the impact of the ASTM International (formerly American Society of Testing Materials) safety standard and associated product safety changes on accidental exposures to liquid laundry packets (LLPs) in children. METHODS: The National Poison Data System was queried for reports of accidental exposures to LLPs in children <6 years old received from 01 July 2012 to 31 December 2018. In 2014, ASTM International began developing a standard specifying voluntary product changes to reduce the risk of LLP exposures in young children. Product changes were made between 2013 and 2016. Exposures were grouped into baseline, transition, and post periods based on the timing of the standard's implementation. Exposure counts and sales adjusted rates were compared between the baseline and post period for all exposures and exposures involving healthcare facility (HCF) evaluation, HCF admission, and major medical outcomes. RESULTS: A total of 73,942 accidental exposures in children <6 years old were reported (baseline: 10,229, 13.8%; transition: 43,507, 58.8%; post: 20,206, 27.3%). The percentage of exposures involving HCF evaluation (41.5% to 33.8%), HCF admission (4.5% to 1.9%), and major medical outcomes (0.6% to 0.1%) decreased from the baseline to post period. Sales adjusted rates of all exposures decreased 57.4% (4.920-2.094 exposures/1 million packets sold). Decreases also occurred in HCF evaluations (65.0% decrease; 2.026-0.708 exposures/1 million packets sold), HCF admissions (81.4% decrease; 0.218-0.041 exposures/1 million packets sold), and major medical outcomes (90.9% decrease; 0.030-0.003 exposures/1 million packets sold). CONCLUSIONS: The morbidity of accidental exposures to LLPs in children decreased substantially following implementation of the ASTM International safety standard. Ongoing monitoring should be performed to determine if additional safety measures are required.

Flow Restrictors and Reduction of Accidental Ingestions of Over-the-Counter Medications.

INTRODUCTION: Flow restrictors are child-resistant packaging innovations designed to limit the amount of liquid dispensed from a medication bottle. In 2011, flow restrictors were added to pediatric liquid single-ingredient acetaminophen formulations. The hypothesis of this study is that implementation would be associated with reduced volume and severity of pediatric acetaminophen exposures reported to the U.S. National Poison Data System. METHODS: This study describes accidental unsupervised ingestions of acetaminophen in children aged <6 years. Exposures were grouped into pre-implementation (pre-period; January 4, 2010-July 17, 2011); transition (July 18, 2011-July 15, 2012); and post-implementation (post-period; July 16, 2012-December 25, 2016) periods. Cumulative and annual rates of change per million units (i.e., bottles) sold were calculated for the pre- and post-periods for acetaminophen and pediatric liquid ibuprofen (comparator without flow restrictors). Pre- to post-period rate ratios were used to compare products and to estimate the potential effect on other over-the-counter medications. Analysis was conducted in 2017 and 2018. RESULTS: The pre- and post-period cumulative acetaminophen exposure rate was 507.2 (95% CI=481.1, 534.6) and 325.6 (95% CI=305.8, 346.7) per 1 million units sold, respectively. Declines in the pre- versus post-period rate ratios were seen for exposures with any effect (0.642, 95% CI=0.591, 0.696) and with clinically significant outcomes (0.728, 95% CI=0.581, 0.913). In the post-period, acetaminophen exposures decreased faster than ibuprofen with a rate of change ratio of 0.936 (95% CI=0.912, 0.960) for all exposures and 0.939 (95% CI=0.909, 0.970) for exposures with any effect. CONCLUSIONS: The addition of flow restrictors to pediatric liquid acetaminophen was associated with a reduction in the number and severity of exposures. Application of flow restrictors to other liquid medications should be considered.