Machine Learning in Relation to Emergency Medicine Clinical and Operational Scenarios: An Overview

Active learning for extracting rare adverse events from electronic health records: A study in pediatric cardiology

OBJECTIVE

Automate the extraction of adverse events from the text of electronic medical records of patients hospitalized for cardiac catheterization.

METHODS

We focused on events related to cardiac catheterization as defined by the NCDR-IMPACT registry. These events were extracted from the Necker Children's Hospital data warehouse. Electronic health records were pre-screened using regular expressions. The resulting datasets contained numerous false positives sentences that were annotated by a cardiologist using an active learning process. A deep learning text classifier was then trained on this active learning-annotated dataset to accurately identify patients who have suffered a serious adverse event.

RESULTS

The dataset included 2,980 patients. Regular expression based extraction of adverse events related to cardiac catheterization achieved a perfect recall. Due to the rarity of adverse events, the dataset obtained from this initial pre-screening step was imbalanced, containing a significant number of false positives. The active learning annotation enabled the acquisition of a representative dataset suitable for training a deep learning model. The deep learning text-classifier identified patients who underwent adverse events after cardiac catheterization with a recall of 0.78 and a specificity of 0.94.

CONCLUSION

Our model effectively identified patients who experienced adverse events related to cardiac catheterization using real clinical data. Enabled by an active learning annotation process, it shows promise for large language model applications in clinical research, especially for rare diseases with limited annotated databases. Our model's strength lies in its development by physicians for physicians, ensuring its relevance and applicability in clinical practice.

Machine learning techniques for prediction in pregnancy complicated by autoimmune rheumatic diseases: Applications and challenges

Autoimmune rheumatic diseases are chronic conditions affecting multiple systems and often occurring in young women of childbearing age. The diseases and the physiological characteristics of pregnancy significantly impact maternal-fetal health and pregnancy outcomes. Currently, the integration of big data with healthcare has led to the increasing popularity of using machine learning (ML) to mine clinical data for studying pregnancy complications. In this review, we introduce the basics of ML and the recent advances and trends of ML in different prediction applications for common pregnancy complications by autoimmune rheumatic diseases. Finally, the challenges and future for enhancing the accuracy, reliability, and clinical applicability of ML in prediction have been discussed. This review will provide insights into the utilization of ML in identifying and assisting clinical decision-making for pregnancy complications, while also establishing a foundation for exploring comprehensive management strategies for pregnancy and enhancing maternal and child health.

Artificial Intelligence in the Intensive Care Unit: Current Evidence on an Inevitable Future Tool

Artificial intelligence (AI) is a technique that attempts to replicate human intelligence, analytical behavior, and decision-making ability. This includes machine learning, which involves the use of algorithms and statistical techniques to enhance the computer's ability to make decisions more accurately. Due to AI's ability to analyze, comprehend, and interpret considerable volumes of data, it has been increasingly used in the field of healthcare. In critical care medicine, where most of the patient load requires timely interventions due to the perilous nature of the condition, AI's ability to monitor, analyze, and predict unfavorable outcomes is an invaluable asset. It can significantly improve timely interventions and prevent unfavorable outcomes, which, otherwise, is not always achievable owing to the constrained human ability to multitask with optimum efficiency. AI has been implicated in intensive care units over the past many years. In addition to its advantageous applications, this article discusses its disadvantages, prospects, and the changes needed to train future critical care professionals. A comprehensive search of electronic databases was performed using relevant keywords. Data from articles pertinent to the topic was assimilated into this review article.

Development and Validation of a Prognostic Nomogram for Patients with AFP and DCP Double-Negative Hepatocellular Carcinoma After Local Ablation

Purpose

Although alpha-fetoprotein (AFP) and des-gamma-carboxyprothrombin (DCP) have a certain predictive ability for the prognosis of hepatocellular carcinoma (HCC), there are still some cases of aggressive recurrence among patients with AFP and DCP double-negative HCC (DNHC) after local ablation. However, prediction models to forecast the prognosis of DNHC patients are still lacking. Thus, this retrospective study aims to explore the prognostic factors in DNHC patients and develop a nomogram to predict recurrence.

Patients and methods

493 DNHC patients who underwent the local ablation at Beijing You'an Hospital between January 1, 2014, and December 31, 2022, were enrolled. A part that was admitted from January 1, 2014, to December 31, 2018, was designated to the training cohort (n = 307); others from January 1, 2019, to December 31, 2022, were allocated to the validation cohort (n = 186). Lasso regression and Cox regression were employed with the aim of screening risk factors and developing the nomogram. The nomogram outcome was assessed by discrimination, calibration, and decision curve analysis (DCA).

Results

Independent prognostic factors selected by Lasso-Cox analysis included age, tumor size, tumor number, and gamma-glutamyl transferase. The area under the receiver operating characteristic (ROC) curves (AUCs) of the training and validation groups (0.738, 0.742, 0.836, and 0.758, 0.821) exhibited the excellent predicted outcome of the nomogram. Calibration plots and DCA plots suggest desirable calibration performance and clinical utility. Patients were stratified into three risk groups by means of the nomogram: low-risk, intermediate-risk, and high-risk, respectively. There exists an obvious distinction in recurrence-free survival (RFS) among three groups (p<0.0001).

Conclusion

In conclusion, we established and validated a nomogram for DNHC patients who received local ablation. The nomogram showed excellent predictive power for the recurrence of HCC and could contribute to guiding clinical decisions.

Natural Language Processing - A Surveillance Stepping Stone to Identify Child Abuse

OBJECTIVE

We aimed to refine a natural language processing (NLP) algorithm that identified injuries associated with child abuse and identify areas in which integration into a real-time clinical decision support (CDS) tool may improve clinical care.

METHODS

We applied an NLP algorithm in "silent mode" to all emergency department (ED) provider notes between July 2021 and December 2022 (n = 353) at 1 pediatric and 8 general EDs. We refined triggers for the NLP, assessed adherence to clinical guidelines, and evaluated disparities in degree of evaluation by examining associations between demographic variables and abuse evaluation or reporting to child protective services.

RESULTS

Seventy-three cases falsely triggered the NLP, often due to errors in interpreting linguistic context. We identified common false-positive scenarios and refined the algorithm to improve NLP specificity. Adherence to recommended evaluation standards for injuries defined by nationally accepted clinical guidelines was 63%. There were significant demographic differences in evaluation and reporting based on presenting ED type, insurance status, and race and ethnicity.

CONCLUSIONS

Analysis of an NLP algorithm in "silent mode" allowed for refinement of the algorithm and highlighted areas in which real-time CDS may help ED providers identify and pursue appropriate evaluation of injuries associated with child physical abuse.

Performance of emergency triage prediction of an open access natural language processing based chatbot application (ChatGPT): A preliminary, scenario-based cross-sectional study

OBJECTIVES

Artificial intelligence companies have been increasing their initiatives recently to improve the results of chatbots, which are software programs that can converse with a human in natural language. The role of chatbots in health care is deemed worthy of research. OpenAI's ChatGPT is a supervised and empowered machine learning-based chatbot. The aim of this study was to determine the performance of ChatGPT in emergency medicine (EM) triage prediction.

METHODS

This was a preliminary, cross-sectional study conducted with case scenarios generated by the researchers based on the emergency severity index (ESI) handbook v4 cases. Two independent EM specialists who were experts in the ESI triage scale determined the triage categories for each case. A third independent EM specialist was consulted as arbiter, if necessary. Consensus results for each case scenario were assumed as the reference triage category. Subsequently, each case scenario was queried with ChatGPT and the answer was recorded as the index triage category. Inconsistent classifications between the ChatGPT and reference category were defined as over-triage (false positive) or under-triage (false negative).

RESULTS

Fifty case scenarios were assessed in the study. Reliability analysis showed a fair agreement between EM specialists and ChatGPT (Cohen's Kappa: 0.341). Eleven cases (22%) were over triaged and 9 (18%) cases were under triaged by ChatGPT. In 9 cases (18%), ChatGPT reported two consecutive triage categories, one of which matched the expert consensus. It had an overall sensitivity of 57.1% (95% confidence interval [CI]: 34-78.2), specificity of 34.5% (95% CI: 17.9-54.3), positive predictive value (PPV) of 38.7% (95% CI: 21.8-57.8), negative predictive value (NPV) of 52.6 (95% CI: 28.9-75.6), and an F1 score of 0.461. In high acuity cases (ESI-1 and ESI-2), ChatGPT showed a sensitivity of 76.2% (95% CI: 52.8-91.8), specificity of 93.1% (95% CI: 77.2-99.2), PPV of 88.9% (95% CI: 65.3-98.6), NPV of 84.4 (95% CI: 67.2-94.7), and an F1 score of 0.821. The receiver operating characteristic curve showed an area under the curve of 0.846 (95% CI: 0.724-0.969, P < 0.001) for high acuity cases.

CONCLUSION

The performance of ChatGPT was best when predicting high acuity cases (ESI-1 and ESI-2). It may be useful when determining the cases requiring critical care. When trained with more medical knowledge, ChatGPT may be more accurate for other triage category predictions.

Development of a Pancreatic Fistula Prediction Model After Pancreaticoduodenectomy Based on a Decision Tree and Random Forest Algorithm

BACKGROUND

The incidence of postoperative pancreatic fistula (POPF) after pancreaticoduodenectomy (PD) is high. We sought to develop a POPF prediction model based on a decision tree (DT) and random forest (RF) algorithm after PD and to explore its clinical value.

METHODS

The case data of 257 patients who underwent PD in a tertiary general hospital from 2013 to 2021 were retrospectively collected in China. The RF model was used to select features by ranking the importance of variables, and both algorithms were used to build the prediction model after automatic adjustment of parameters by setting the respective hyperparameter intervals and resampling as a 10-fold cross-validation method, etc. The prediction model's performance was assessed by the receiver operating characteristic curve (ROC) and the area under curve (AUC).

RESULTS

Postoperative pancreatic fistula occurred in 56 cases (56/257, 21.8%). The DT model had an AUC of .743 and an accuracy of .840, while the RF model had an AUC of .977 and an accuracy of .883. The DT plot visualized the process of inferring the risk of pancreatic fistula from the DT model on independent individuals. The top 10 important variables were selected for ranking in the RF variable importance ranking.

CONCLUSION

This study successfully developed a DT and RF algorithm for the POPF prediction model, which can be used as a reference for clinical health care professionals to optimize treatment strategies to reduce the incidence of POPF.

Machine Learning for Predicting the Development of Postoperative Acute Kidney Injury After Coronary Artery Bypass Grafting Without Extracorporeal Circulation

Background: Cardiac surgery-associated acute kidney injury (CSA-AKI) is a major complication that increases morbidity and mortality after cardiac surgery. Most established predictive models are limited to the analysis of nonlinear relationships and do not adequately consider intraoperative variables and early postoperative variables. Nonextracorporeal circulation coronary artery bypass grafting (off-pump CABG) remains the procedure of choice for most coronary surgeries, and refined CSA-AKI predictive models for off-pump CABG are notably lacking. Therefore, this study used an artificial intelligence-based machine learning approach to predict CSA-AKI from comprehensive perioperative data.

Methods: In total, 293 variables were analysed in the clinical data of patients undergoing off-pump CABG in the Department of Cardiac Surgery at the First Affiliated Hospital of Guangxi Medical University between 2012 and 2021. According to the KDIGO criteria, postoperative AKI was defined by an elevation of at least 50% within 7 days, or 0.3 mg/dL within 48 hours, with respect to the reference serum creatinine level. Five machine learning algorithms—a simple decision tree, random forest, support vector machine, extreme gradient boosting and gradient boosting decision tree (GBDT)—were used to construct the CSA-AKI predictive model. The performance of these models was evaluated with the area under the receiver operating characteristic curve (AUC). Shapley additive explanation (SHAP) values were used to explain the predictive model.

Results: The three most influential features in the importance matrix plot were 1-day postoperative serum potassium concentration, 1-day postoperative serum magnesium ion concentration, and 1-day postoperative serum creatine phosphokinase concentration.

Conclusion: GBDT exhibited the largest AUC (0.87) and can be used to predict the risk of AKI development after surgery, thus enabling clinicians to optimise treatment strategies and minimise postoperative complications.

Effect of Applying a Real-Time Medical Record Input Assistance System With Voice Artificial Intelligence on Triage Task Performance in the Emergency Department: Prospective Interventional Study

BACKGROUND

Natural language processing has been established as an important tool when using unstructured text data; however, most studies in the medical field have been limited to a retrospective analysis of text entered manually by humans. Little research has focused on applying natural language processing to the conversion of raw voice data generated in the clinical field into text using speech-to-text algorithms.

OBJECTIVE

In this study, we investigated the promptness and reliability of a real-time medical record input assistance system with voice artificial intelligence (RMIS-AI) and compared it to the manual method for triage tasks in the emergency department.

METHODS

From June 4, 2021, to September 12, 2021, RMIS-AI, using a machine learning engine trained with 1717 triage cases over 6 months, was prospectively applied in clinical practice in a triage unit. We analyzed a total of 1063 triage tasks performed by 19 triage nurses who agreed to participate. The primary outcome was the time for participants to perform the triage task.

RESULTS

The median time for participants to perform the triage task was 204 (IQR 155, 277) seconds by RMIS-AI and 231 (IQR 180, 313) seconds using manual method; this difference was statistically significant (P<.001). Most variables required for entry in the triage note showed a higher record completion rate by the manual method, but in the recording of additional chief concerns and past medical history, RMIS-AI showed a higher record completion rate than the manual method. Categorical variables entered by RMIS-AI showed less accuracy compared with continuous variables, such as vital signs.

CONCLUSIONS

RMIS-AI improves the promptness in performing triage tasks as compared to using the manual input method. However, to make it a reliable alternative to the conventional method, technical supplementation and additional research should be pursued.

Development and Validation of a Natural Language Processing Tool to Identify Injuries in Infants Associated With Abuse

OBJECTIVES

Medically minor but clinically important findings associated with physical child abuse, such as bruises in pre-mobile infants, may be identified by frontline clinicians yet the association of these injuries with child abuse is often not recognized, potentially allowing the abuse to continue and even to escalate. An accurate natural language processing (NLP) algorithm to identify high-risk injuries in electronic health record notes could improve detection and awareness of abuse. The objectives were to: 1) develop an NLP algorithm that accurately identifies injuries in infants associated with abuse and 2) determine the accuracy of this algorithm.

METHODS

An NLP algorithm was designed to identify ten specific injuries known to be associated with physical abuse in infants. Iterative cycles of review identified inaccurate triggers, and coding of the algorithm was adjusted. The optimized NLP algorithm was applied to emergency department (ED) providers' notes on 1344 consecutive sample of infants seen in 9 EDs over 3.5 months. Results were compared with review of the same notes conducted by a trained reviewer blind to the NLP results with discrepancies adjudicated by a child abuse expert.

RESULTS

Among the 1344 encounters, 41 (3.1%) had one of the high-risk injuries. The NLP algorithm had a sensitivity and specificity of 92.7% (95% confidence interval [CI]: 79.0%-98.1%) and 98.1% (95% CI: 97.1%-98.7%), respectively, and positive and negative predictive values were 60.3% and 99.8%, respectively, for identifying high-risk injuries.

CONCLUSIONS

An NLP algorithm to identify infants with high-risk injuries in EDs has good accuracy and may be useful to aid clinicians in the identification of infants with injuries associated with child abuse.

Using machine learning to predict subsequent events after EMS non-conveyance decisions

Background

Predictors of subsequent events after Emergency Medical Services (EMS) non-conveyance decisions are still unclear, though patient safety is the priority in prehospital emergency care. The aim of this study was to find out whether machine learning can be used in this context and to identify the predictors of subsequent events based on narrative texts of electronic patient care records (ePCR).

Methods

This was a prospective cohort study of EMS patients in Finland. The data was collected from three different regions between June 1 and November 30, 2018. Machine learning, in form of text classification, and manual evaluation were used to predict subsequent events from the clinical notes after a non-conveyance mission.

Results

FastText-model (AUC 0.654) performed best in prediction of subsequent events after EMS non-conveyance missions (n = 11,846). The model and manual analyses showed that many of the subsequent events were planned before, EMS guided the patients to visit primary health care facilities or ED next or following days after non-conveyance. The most frequent signs and symptoms as subsequent event predictors were musculoskeletal-, infection-related and non-specific complaints. 1 in 5 the EMS documentation was inadequate and many of these led to a subsequent event.

Conclusion

Machine learning can be used to predict subsequent events after EMS non-conveyance missions. From the patient safety perspective, it is notable that subsequent event does not necessarily mean that patient safety is compromised. There were a number of subsequent visits to primary health care or EDs, which were planned before by EMS. This demonstrates the appropriate use of limited resources to avoid unnecessary conveyance to the ED. However, further studies are needed without planned subsequent events to find out the harmful subsequent events, where EMS non-conveyance puts patient safety at risk.

Artificial intelligence and machine learning in emergency medicine: a narrative review

AIM

The emergence and evolution of artificial intelligence (AI) has generated increasing interest in machine learning applications for health care. Specifically, researchers are grasping the potential of machine learning solutions to enhance the quality of care in emergency medicine.

METHODS

We undertook a narrative review of published works on machine learning applications in emergency medicine and provide a synopsis of recent developments.

RESULTS

This review describes fundamental concepts of machine learning and presents clinical applications for triage, risk stratification specific to disease, medical imaging, and emergency department operations. Additionally, we consider how machine learning models could contribute to the improvement of causal inference in medicine, and to conclude, we discuss barriers to safe implementation of AI.

CONCLUSION

We intend that this review serves as an introduction to AI and machine learning in emergency medicine.

Racial underrepresentation in dermatological datasets leads to biased machine learning models and inequitable healthcare

Objective

Clinical applications of machine learning are promising as a tool to improve patient outcomes through assisting diagnoses, treatment, and analyzing risk factors for screening. Possible clinical applications are especially prominent in dermatology as many diseases and conditions present visually. This allows a machine learning model to analyze and diagnose conditions using patient images and data from electronic health records (EHRs) after training on clinical datasets but could also introduce bias. Despite promising applications, artificial intelligence has the capacity to exacerbate existing demographic disparities in healthcare if models are trained on biased datasets.

Methods

Through systematic literature review of available literature, we highlight the extent of bias present in clinical datasets as well as the implications it could have on healthcare if not addressed.

Results

We find the implications are worsened in dermatological models. Despite the severity and complexity of melanoma and other dermatological diseases as well as differing disease presentations based on skin-color, many imaging datasets underrepresent certain demographic groups causing machine learning models to train on images of primarily fair-skinned individuals leaving minorities behind.

Conclusion

In order to address this disparity, research first needs to be done investigating the extent of the bias present and the implications it may have on equitable healthcare.

Machine learning approach to predict acute kidney injury after liver surgery

BACKGROUND

Acute kidney injury (AKI) after surgery appears to increase the risk of death in patients with liver cancer. In recent years, machine learning algorithms have been shown to offer higher discriminative efficiency than classical statistical analysis.

AIM

To develop prediction models for AKI after liver cancer resection using machine learning techniques.

METHODS

We screened a total of 2450 patients who had undergone primary hepatocellular carcinoma resection at Changzheng Hospital, Shanghai City, China, from January 1, 2015 to August 31, 2020. The AKI definition used was consistent with the Kidney Disease: Improving Global Outcomes. We included in our analysis preoperative data such as demographic characteristics, laboratory findings, comorbidities, and medication, as well as perioperative data such as duration of surgery. Computerized algorithms used for model development included logistic regression (LR), support vector machine (SVM), random forest (RF), extreme gradient boosting (XGboost), and decision tree (DT). Feature importance was also ranked according to its contribution to model development.

RESULTS

AKI events occurred in 296 patients (12.1%) within 7 d after surgery. Among the original models based on machine learning techniques, the RF algorithm had optimal discrimination with an area under the curve value of 0.92, compared to 0.87 for XGBoost, 0.90 for DT, 0.90 for SVM, and 0.85 for LR. The RF algorithm also had the highest concordance-index (0.86) and the lowest Brier score (0.076). The variable that contributed the most in the RF algorithm was age, followed by cholesterol, and surgery time.

CONCLUSION

Machine learning algorithms are highly effective in discriminating patients at high risk of developing AKI. The successful application of machine learning models may help guide clinical decisions and help improve the long-term prognosis of patients.

Automated and accurate assessment for postural abnormalities in patients with Parkinson's disease based on Kinect and machine learning

Background

Automated and accurate assessment for postural abnormalities is necessary to monitor the clinical progress of Parkinson’s disease (PD). The combination of depth camera and machine learning makes this purpose possible.

Methods

Kinect was used to collect the postural images from 70 PD patients. The collected images were processed to extract three-dimensional body joints, which were then converted to two-dimensional body joints to obtain eight quantified coronal and sagittal features (F1-F8) of the trunk. The decision tree classifier was carried out over a data set established by the collected features and the corresponding doctors’ MDS-UPDRS-III 3.13 (the 13th item of the third part of Movement Disorder Society-Sponsored Revision of the Unified Parkinson’s Disease Rating Scale) scores. An objective function was implanted to further improve the human–machine consistency.

Results

The automated grading of postural abnormalities for PD patients was realized with only six selected features. The intraclass correlation coefficient (ICC) between the machine’s and doctors’ score was 0.940 (95%CI, 0.905–0.962), meaning the machine was highly consistent with the doctors’ judgement. Besides, the decision tree classifier performed outstandingly, reaching 90.0% of accuracy, 95.7% of specificity and 89.1% of sensitivity in rating postural severity.

Conclusions

We developed an intelligent evaluation system to provide accurate and automated assessment of trunk postural abnormalities in PD patients. This study demonstrates the practicability of our proposed method in the clinical scenario to help making the medical decision about PD.

Ultrasound Images Guided under Deep Learning in the Anesthesia Effect of the Regional Nerve Block on Scapular Fracture Surgery

In order to discuss the clinical characteristics of patients with scapular fracture, deep learning model was adopted in ultrasound images of patients to locate the anesthesia point of patients during scapular fracture surgery treated with the regional nerve block. 100 patients with scapular fracture who were hospitalized for emergency treatment in the hospital were recruited. Patients in the algorithm group used ultrasound-guided regional nerve block puncture, and patients in the control group used traditional body surface anatomy for anesthesia positioning. The ultrasound images of the scapula of the contrast group were used for the identification of the deep learning model and analysis of anesthesia acupuncture sites. The ultrasound images of the scapula anatomy of the patients in the contrast group were extracted, and the convolutional neural network model was employed for training and test. Moreover, the model performance was evaluated. It was found that the adoption of deep learning greatly improved the accuracy of the image. It took an average of 7.5 ± 2.07 minutes from the time the puncture needle touched the skin to the completion of the injection in the algorithm group (treated with artificial intelligence ultrasound positioning). The operation time of the control group (anatomical positioning) averaged 10.2 ± 2.62 min. Moreover, there was a significant difference between the two groups (p < 0.05). The method adopted in the contrast group had high positioning accuracy and good anesthesia effect, and the patients had reduced postoperative complications of patients (all P < 0.005). The deep learning model can effectively improve the accuracy of ultrasound images and measure and assist the treatment of future clinical cases of scapular fractures. While improving medical efficiency, it can also accurately identify patient fractures, which has great adoption potential in improving the effect of surgical anesthesia.

Machine Learning and Precision Medicine in Emergency Medicine: The Basics

As machine learning (ML) and precision medicine become more readily available and used in practice, emergency physicians must understand the potential advantages and limitations of the technology. This narrative review focuses on the key components of machine learning, artificial intelligence, and precision medicine in emergency medicine (EM). Based on the content expertise, we identified articles from EM literature. The authors provided a narrative summary of each piece of literature. Next, the authors provided an introduction of the concepts of ML, artificial intelligence as an extension of ML, and precision medicine. This was followed by concrete examples of their applications in practice and research. Subsequently, we shared our thoughts on how to consume the existing research in these subjects and conduct high-quality research for academic emergency medicine. We foresee that the EM community will continue to adapt machine learning, artificial intelligence, and precision medicine in research and practice. We described several key components using our expertise.

Development and Validation of Machine Learning Models to Predict Admission From Emergency Department to Inpatient and Intensive Care Units

STUDY OBJECTIVE

This study aimed to develop and validate 2 machine learning models that use historical and current-visit patient data from electronic health records to predict the probability of patient admission to either an inpatient unit or ICU at each hour (up to 24 hours) of an emergency department (ED) encounter. The secondary goal was to provide a framework for the operational implementation of these machine learning models.

METHODS

Data were curated from 468,167 adult patient encounters in 3 EDs (1 academic and 2 community-based EDs) of a large academic health system from August 1, 2015, to October 31, 2018. The models were validated using encounter data from January 1, 2019, to December 31, 2019. An operational user dashboard was developed, and the models were run on real-time encounter data.

RESULTS

For the intermediate admission model, the area under the receiver operating characteristic curve was 0.873 and the area under the precision-recall curve was 0.636. For the ICU admission model, the area under the receiver operating characteristic curve was 0.951 and the area under the precision-recall curve was 0.461. The models had similar performance in both the academic- and community-based settings as well as across the 2019 and real-time encounter data.

CONCLUSION

Machine learning models were developed to accurately make predictions regarding the probability of inpatient or ICU admission throughout the entire duration of a patient's encounter in ED and not just at the time of triage. These models remained accurate for a patient cohort beyond the time period of the initial training data and were integrated to run on live electronic health record data, with similar performance.

Impact of Big Data Analytics on People's Health: Overview of Systematic Reviews and Recommendations for Future Studies

Background

Although the potential of big data analytics for health care is well recognized, evidence is lacking on its effects on public health.

Objective

The aim of this study was to assess the impact of the use of big data analytics on people’s health based on the health indicators and core priorities in the World Health Organization (WHO) General Programme of Work 2019/2023 and the European Programme of Work (EPW), approved and adopted by its Member States, in addition to SARS-CoV-2–related studies. Furthermore, we sought to identify the most relevant challenges and opportunities of these tools with respect to people’s health.

Methods

Six databases (MEDLINE, Embase, Cochrane Database of Systematic Reviews via Cochrane Library, Web of Science, Scopus, and Epistemonikos) were searched from the inception date to September 21, 2020. Systematic reviews assessing the effects of big data analytics on health indicators were included. Two authors independently performed screening, selection, data extraction, and quality assessment using the AMSTAR-2 (A Measurement Tool to Assess Systematic Reviews 2) checklist.

Results

The literature search initially yielded 185 records, 35 of which met the inclusion criteria, involving more than 5,000,000 patients. Most of the included studies used patient data collected from electronic health records, hospital information systems, private patient databases, and imaging datasets, and involved the use of big data analytics for noncommunicable diseases. “Probability of dying from any of cardiovascular, cancer, diabetes or chronic renal disease” and “suicide mortality rate” were the most commonly assessed health indicators and core priorities within the WHO General Programme of Work 2019/2023 and the EPW 2020/2025. Big data analytics have shown moderate to high accuracy for the diagnosis and prediction of complications of diabetes mellitus as well as for the diagnosis and classification of mental disorders; prediction of suicide attempts and behaviors; and the diagnosis, treatment, and prediction of important clinical outcomes of several chronic diseases. Confidence in the results was rated as “critically low” for 25 reviews, as “low” for 7 reviews, and as “moderate” for 3 reviews. The most frequently identified challenges were establishment of a well-designed and structured data source, and a secure, transparent, and standardized database for patient data.

Conclusions

Although the overall quality of included studies was limited, big data analytics has shown moderate to high accuracy for the diagnosis of certain diseases, improvement in managing chronic diseases, and support for prompt and real-time analyses of large sets of varied input data to diagnose and predict disease outcomes.

Trial Registration

International Prospective Register of Systematic Reviews (PROSPERO) CRD42020214048; https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=214048

High-precision multiclass cell classification by supervised machine learning on lectin microarray data

INTRODUCTION

Establishment of a cell classification platform for evaluation and selection of human pluripotent stem cells (hPSCs) is of great importance to assure the efficacy and safety of cell-based therapy. In our previous work, we introduced a discriminant function that evaluates pluripotency from the cells' glycome. However, it is not yet suitable for general use.

METHODS

The current study aims to establish a high-precision cell classification platform introducing supervised machine learning and test the platform on glycome analysis as a proof-of-concept study. We employed linear classification and neural network to the lectin microarray data from 1577 human cells and categorized them into five classes including hPSCs.

RESULTS

The linear-classification-based model and the neural-network-based model successfully predicted the sample type with accuracies of 89% and 97%, respectively.

CONCLUSIONS

Because of the high recognition accuracies and the small amount of computing resources required for these analyses, our platform can be a high precision conventional cell classification system for hPSCs.

Forty years of emergency medicine research: Uncovering research themes and trends through topic modeling

STUDY OBJECTIVE

Topic identification can facilitate knowledge curation, discover thematic relationships, trends, and predict future direction. We aimed to determine through an unsupervised, machine learning approach to topic modeling the most common research themes in emergency medicine over the last 40 years and summarize their trends and characteristics.

METHODS

We retrieved the complete reference entries including article abstracts from Ovid for all original research articles from 1980 to 2019 within emergency medicine for six widely-cited journals. Abstracts were processed through a natural language pipeline and analyzed by a latent Dirichlet allocation topic modeling algorithm for unsupervised topic discovery. Topics were further examined through trend analysis, word associations, co-occurrence metrics, and two-dimensional embeddings.

RESULTS

We retrieved 47,158 articles during the defined time period that were filtered to 20,528 articles for further analysis. Forty topics covering methodologic and clinical areas were discovered. These topics separated into distinct clusters when embedded in two-dimensional space and exhibited consistent patterns of interaction. We observed the greatest increase in popularity in research themes involving risk factors (0.4% to 5.2%), health utilization (1.2% to 5.0%), and ultrasound (0.7% to 3.3%), and a relative decline in research involving basic science (8.9% to 1.1%), cardiac arrest (6.5% to 2.2%), and vitals (6.3% to 1.3%) over the past 40 years. Our data show only very modest growth in mental health and substance abuse research (1.0% to 1.6%), despite ongoing crises.

CONCLUSIONS

Topic modeling via unsupervised machine learning applied to emergency medicine abstracts discovered coherent topics, trends, and patterns of interaction.

Predicting Acute Kidney Injury after Cardiac Surgery by Machine Learning Approaches

Cardiac surgery-associated AKI (CSA-AKI) is common after cardiac surgery and has an adverse impact on short- and long-term mortality. Early identification of patients at high risk of CSA-AKI by applying risk prediction models allows clinicians to closely monitor these patients and initiate effective preventive and therapeutic approaches to lessen the incidence of AKI. Several risk prediction models and risk assessment scores have been developed for CSA-AKI. However, the definition of AKI and the variables utilized in these risk scores differ, making general utility complex. Recently, the utility of artificial intelligence coupled with machine learning, has generated much interest and many studies in clinical medicine, including CSA-AKI. In this article, we discussed the evolution of models established by machine learning approaches to predict CSA-AKI.

Predicting outcome for ambulance patients with dyspnea: a prospective cohort study

OBJECTIVE

To validate the discrimination and classification accuracy of a novel acute dyspnea scale for identifying outcomes of out-of-hospital patients with acute dyspnea.

METHODS

Prospective observational population-based study in the North Denmark Region. We included patients from July 1, 2017 to September 24, 2019 assessed as having acute dyspnea by the emergency dispatcher or by emergency medical services (EMS) personnel. Patients rated dyspnea using the 11-point acute dyspnea scale. The primary outcomes were hospitalization >2 days, ICU admission within 48 hours of ambulance run, and 30-day mortality. We used 5-fold cross-validation and area under receiver operating curves (AUC) to assess predictive properties of the acute dyspnea scale score alone and combined with vital data, age, and sex.

RESULTS

We included 3144 EMS patients with reported dyspnea. Median acute dyspnea scale score was 7 (interquartile range 5 to 8). The outcomes were: 1966 (63%) hospitalized, 164 (5%) ICU stay, and 224 (9%) died within 30 days of calling the ambulance. The acute dyspnea scale score alone showed poor discrimination for hospitalization (AUC 0.56, 95% confidence intervals: 0.54-0.58), intensive care unit admission (0.58, 0.53-0.62), and mortality (0.46, 0.41-0.50). Vital signs (respiratory rate, blood oxygen saturation, blood pressure, and heart rate) showed similarly poor discrimination for all outcomes. The combination of [vital signs + acute dyspnea scale score] showed better discrimination for hospitalization, ICU admission, and mortality (AUC 0.71-0.72). Patients not able to report an acute dyspnea scale score worse outcomes on all parameters.

CONCLUSION

The dyspnea scale showed poor accuracy and discrimination when predicting hospitalization, stay at intensive care unit, and mortality on its own. However, the dyspnea scale may be beneficial as performance measure and indicator of out-of-hospital care.