Journal of Medical Internet Research - International Scientific Journal for Medical Research, Information and Communication on the Internet

Predicting the risk of falls in advance can benefit the quality of care and potentially reduce mortality and morbidity in the older population. The aim of this study was to construct and validate an electronic health record-based fall risk predictive tool to identify elders at a higher risk of falls. The one-year fall prediction model was developed using the machine-learning-based algorithm, XGBoost, and tested on an independent validation cohort. The data were collected from electronic health records (EHR) of Maine from 2016 to 2018, comprising 265,225 older patients (≥65 years of age). This model attained a validated C-statistic of 0.807, where 50 % of the identified high-risk true positives were confirmed to fall during the first 94 days of next year. The model also captured in advance 58.01 % and 54.93 % of falls that happened within the first 30 and 30-60 days of next year. The identified high-risk patients of fall showed conditions of severe disease comorbidities, an enrichment of fall-increasing cardiovascular and mental medication prescriptions and increased historical clinical utilization, revealing the complexity of the underlying fall etiology. The XGBoost algorithm captured 157 impactful predictors into the final predictive model, where cognitive disorders, abnormalities of gait and balance, Parkinson's disease, fall history and osteoporosis were identified as the top-5 strongest predictors of the future fall event. By using the EHR data, this risk assessment tool attained an improved discriminative ability and can be immediately deployed in the health system to provide automatic early warnings to older adults with increased fall risk and identify their personalized risk factors to facilitate customized fall interventions.

Journal of Medical Internet Research - International Scientific Journal for Medical Research, Information and Communication on the Internet #Preprint #PeerReviewMe: Warning: This is a unreviewed preprint. Readers are warned that the document has not been peer-reviewed by expert/patient reviewers or an academic editor, may contain misleading claims, and is likely to undergo changes before final publication, if accepted, or may have been rejected/withdrawn. Readers with interest and expertise are encouraged to sign up as peer-reviewer, if the paper is within an open peer-review period. Please cite this preprint only for review purposes or for grant applications and CVs (if you are the author). Background: Some hospitalized patients rapidly deteriorate due to either disease progression or imperfect triage and level of care assignment after their admission. An EWS to identify patients at high risk of subsequent intra-hospital death can be an effective tool to benefit patient safety and quality of care, and also reduce avoidable harm and costs. Objective: To prospectively validate a real-time early warning system (EWS) with the capacity to predict patients at high risk of in-hospital mortality during their inpatient episodes. Methods: Data were collected from the system-wide electronic medical record (EMR) of two acute Berkshire Health System hospitals, comprised of 54,246 inpatient admissions from January 01, 2015 to September 30, 2017, of which 3.97% resulted in intra-hospital deaths. After constructing the model using statistical methods and modern machine learning algorithms, we prospectively validated the algorithms as a real-time inpatient early warning system (EWS) for mortality. Results: Our EWS algorithm scored patients’ daily and long-term risk of in-hospital mortality probability after admission and stratified them into distinct risk groups. In the prospective validation, our EWS prospectively attained a c-statistic of 0.884, where 99 encounters were captured in the highest risk group, 68.69% of which died during the episodes. Furthermore, our real-time EWS successfully forecasted the top 13.33% of the expired patients at least 40.8 hours before death. Important clinical utilization features, together with coded diagnoses, vital signs and laboratory test results were recognized as impactful predictors in the final EWS. Conclusions: In this study, we prospectively demonstrated our new EWS’s capability of real-time monitoring and alerting clinicians to patients at high risk of in-hospital death, providing opportunities for timely interventions. This real-time EWS is able to assist clinical decision making and enable more actionable and effective individualized care for better patients’ health outcomes in target medical facilities.

Journal of Medical Internet Research - International Scientific Journal for Medical Research, Information and Communication on the Internet #Preprint #PeerReviewMe: Warning: This is a unreviewed preprint. Readers are warned that the document has not been peer-reviewed by expert/patient reviewers or an academic editor, may contain misleading claims, and is likely to undergo changes before final publication, if accepted, or may have been rejected/withdrawn. Readers with interest and expertise are encouraged to sign up as peer-reviewer, if the paper is within an open peer-review period. Please cite this preprint only for review purposes or for grant applications and CVs (if you are the author). Background: Lung cancer is the leading cause of cancer death worldwide. Early detection of individuals at risk of lung cancer is critical to reduce the mortality rate. Objective: The aim of this study was to develop and validate a prospective risk prediction model to identify patients at risk of receiving a new diagnosis of lung cancer (incident lung cancer) within the next one year in a general population. Methods: Data from individual patient electronic health records (EHRs) were extracted from the Maine Health Information Exchange network. The study population consisted of patients having at least one EHR between April 1, 2016 and March 31, 2018, who had no history of lung cancer. A retrospective (N=873,598) cohort and a prospective (N=836,659) cohort were formed for model construction and validation. An Extreme Gradient Boosting (XGBoost) algorithm was adopted to build the model. It assigned a score to each individual that measured the probability of receiving a new diagnosis of lung cancer from October 1, 2016 to September 31, 2017, based on the clinical profile in the retrospective cohort from the preceding 6 months. The model was validated using the prospective cohort to predict risk of incident lung cancer from April 1, 2017 to March 31, 2018, using the preceding 6 months’ clinical profile. Results: The model had an area under the curve (AUC) of 0.881(95% CI 0.873-0.889) in the prospective cohort. Two thresholds of 0.0045 and 0.01 were applied to the predictive scores to stratify the population into low-, medium- and high-risk categories. The incidence of lung cancer in the high-risk category (579/53,922, 1.07%) was 7.7 times higher than the incidence in the overall cohort (1,167/836,659, 0.14%). Age, a history of pulmonary diseases and other chronic diseases, medications for mental disorders, and social disparities were found to be associated with incident lung cancer. Conclusions: We developed and prospectively validated an accurate risk prediction model of incident lung cancer in the next one year by statistically learning from the statewide EHR data in the preceding 6 months. Our model was able to identify patients at high risk of incident lung cancer, which will benefit population health through preventive interventions or more intensive surveillance.

Necrotizing Enterocolitis (NEC) is associated with prematurity, enteral feedings, and enteral dysbiosis. Accordingly, we hypothesized that along with nutritional variability, metabolic dysfunction would be associated with NEC onset. Methods: We queried a multicenter longitudinal database that included 995 preterm infants (<32 weeks gestation) and included 73 cases of NEC. Dried blood spot samples were obtained on day of life 1, 7, 28, and 42. Metabolite data from each time point included 72 amino acid (AA) and acylcarnitine (AC) measures. Nutrition data were averaged at each of the same time points. Odds ratios and 95% confidence intervals were calculated using samples obtained prior to NEC diagnosis and adjusted for potential confounding variables. Nutritional and metabolic data were plotted longitudinally to determine relationship to NEC onset. Results: Day 1 analyte levels of alanine, phenylalanine, free carnitine, C16, arginine, C14:1/C16, and citrulline/phenylalanine were associated with the subsequent development of NEC. Over time, differences in individual analyte levels associated with NEC onset shifted from predominantly AAs at birth to predominantly ACs by day 42. Subjects who developed NEC received significantly lower weight-adjusted total calories (p < 0.001) overall, a trend that emerged by day of life 7 (p = 0.020), and persisted until day of life 28 (p < 0.001) and 42 (p < 0.001). Conclusion: Premature infants demonstrate metabolic differences at birth. Metabolite abnormalities progress in parallel to significant differences in nutritional delivery signifying metabolic dysfunction in premature newborns prior to NEC onset. These observations provide new insights to potential contributing pathophysiology of NEC and opportunity for clinical care-based prevention.

Backgrounds: Early and accurate diagnosis of pediatric pneumonia in primary health care can reduce the chance of long-term respiratory diseases, related hospitalizations and mortality while lowering medical costs. The aim of this study was to assess the value of blood biomarkers, clinical symptoms and their combination in assisting discrimination of pneumonia from upper respiratory tract infection (URTI) in children. Methods: Both univariate and multivariate logistic regressions were used to build the pneumonia screening model based on a retrospective cohort, comprised of 5211 children (age ≤ 18 years). The electronic health records of the patients, who had inpatient admission or outpatient visits between February 15, 2012 to September 30, 2018, were extracted from the hospital information system of Zhejiang Provincial People’s Hospital, Hangzhou, Zhejiang Province, China. The children who were diagnosed with pneumonia and URTI were enrolled and their clinical features and levels of blood biomarkers were compared. Using the area under the ROC curve, both two screening models were evaluated under 80% (training) versus 20% (test) cross-validation data split for their accuracy. Results: In the retrospective cohort, 2548 of 5211 children were diagnosed with the defined pneumonia. The univariate screening model reached predicted AUCs of 0.76 for lymphocyte/monocyte ratio (LMR) and 0.71 for neutrophil/lymphocyte ratio (NLR) when identified overall pneumonia from URTI, attaining the best performance among the biomarker candidates. In subgroup analysis, LMR and NLR attained AUCs of 0.80 and 0.86 to differentiate viral pneumonia from URTI, and AUCs of 0.77 and 0.71 to discriminate bacterial pneumonia from URTI respectively. After integrating LMR and NLR with three clinical symptoms of fever, cough and rhinorrhea, the multivariate screening model obtained increased predictive values, reaching validated AUCs of 0.84, 0.95 and 0.86 for distinguishing pneumonia, viral pneumonia and bacterial pneumonia from URTI respectively. Conclusions: Our study demonstrated that combining LMR and NLR with critical clinical characteristics reached promising accuracy in differentiating pneumonia from URTI, thus could be considered as a useful screening tool to assist the diagnosis of pneumonia, in particular, in community healthcare centers. Further researches could be conducted to evaluate the model’s clinical utility and cost-effectiveness in primary care scenarios to facilitate pneumonia diagnosis, especially in rural settings.