A tree-based decision model to support prediction of the severity of asthma exacerbations in children

A scoping review of asthma and machine learning

OBJECTIVE

The objective of this study was to determine the extent of machine learning (ML) application in asthma research and to identify research gaps while mapping the existing literature.

DATA SOURCES

We conducted a scoping review. PubMed, ProQuest, and Embase Scopus databases were searched with an end date of September 18, 2020.

STUDY SELECTION

DistillerSR was used for data management. Inclusion criteria were an asthma focus, human participants, ML techniques, and written in English. Exclusion criteria were abstract only, simulation-based, not human based, or were reviews or commentaries. Descriptive statistics were presented.

RESULTS

A total of 6,317 potential articles were found. After removing duplicates, and reviewing the titles and abstracts, 102 articles were included for the full text analysis. Asthma episode prediction (24.5%), asthma phenotype classification (16.7%), and genetic profiling of asthma (12.7%) were the top three study topics. Cohort (52.9%), cross-sectional (20.6%), and case-control studies (11.8%) were the study designs most frequently used. Regarding the ML techniques, 34.3% of the studies used more than one technique. Neural networks, clustering, and random forests were the most common ML techniques used where they were used in 20.6%, 18.6%, and 17.6% of studies, respectively. Very few studies considered location of residence (i.e. urban or rural status).

CONCLUSIONS

The use of ML in asthma studies has been increasing with most of this focused on the three major topics (>50%). Future research using ML could focus on gaps such as a broader range of study topics and focus on its use in additional populations (e.g. location of residence).

Supplemental data for this article is available online at http://dx.doi.org/ .

Risk factors for pressure ulcer recurrence following surgical reconstruction: A cross-sectional retrospective analysis

Many studies on the recurrence of pressure ulcers after surgical reconstruction have focused on surgical techniques and socioeconomic factors. Herein, we aimed to identify the risk factors of the associated comorbidities for pressure ulcer recurrence. We enrolled 147 patients who underwent pressure ulcer reconstruction and were followed up for more than three years. The recurrence of pressure ulcers was defined as recurrent pressure ulcers with stage 3/4 pressure ulcers. We reviewed and analyzed systematic records of medical histories, including sex, age, associated comorbidities such as spinal cord injury (SCI), diabetes mellitus (DM), coronary artery disease, cerebral vascular accident, end-stage renal disease, scoliosis, dementia, Parkinson's disease, psychosis, autoimmune diseases, hip surgery, and locations of the primary pressure ulcer. Patients with recurrent pressure ulcers were younger than those without. Patients with SCI and scoliosis had higher odds, while those with Parkinson's disease had lower odds of recurrence of pressure ulcers than those without these comorbidities. Moreover, the decision tree algorithm identified that SCI, DM, and age < 34 years could be risk factor classifiers for predicting recurrent pressure ulcers. This study demonstrated that age and SCI are the two most important risk factors associated with recurrent pressure ulcers following surgical reconstruction.

Decision tree-based rules outperform risk scores for childhood asthma prognosis

BACKGROUND

There are no widely accepted prognostic tools for childhood asthma; this is in part due to the multifactorial and time-dependent nature of mechanisms and risk factors that contribute to asthma development. Our study objective was to develop and evaluate the prognostic performance of conditional inference decision tree-based rules using the Pediatric Asthma Risk Score (PARS) predictors as an alternative to the existing logistic regression-based risk score for childhood asthma prediction at 7 years in a high-risk population.

METHODS

The Canadian Asthma Primary Prevention Study data were used to develop, compare, and contrast the prognostic performance (area under the curve [AUC], sensitivity, and specificity) of conditional inference tree-based decision rules to the pediatric asthma risk score for the prediction of childhood asthma at 7 years.

RESULTS

Conditional inference decision tree-based rules have higher prognostic performance (AUC: 0.85; 95% CI: 0.81, 0.88; sensitivity = 47%; specificity = 93%) than the pediatric asthma risk score at an optimal cutoff of ≥6 (AUC: 0.71; 95% CI: 0.67, 0.76; sensitivity = 60%; specificity = 74%). Moreover, the pediatric asthma risk score is not linearly related to asthma risk, and at any given pediatric asthma risk score value, different combinations of its pediatric asthma risk score clinical variables differentially predict asthma risk.

CONCLUSION

Conditional inference tree-based decision rules could be a useful childhood asthma prognostic tool, providing an alternative way to identify unique subgroups of at-risk children, and insights into associations and effect mechanisms that are suggestive of appropriate tailored preventive interventions. However, the feasibility and effectiveness of such decision rules in clinical practice is warranted.

Predicting hospitalization of pediatric asthma patients in emergency departments using machine learning

MOTIVATION

The timely identification of patients for hospitalization in emergency departments (EDs) can facilitate efficient use of hospital resources. Machine learning can help the early prediction of ED disposition; however, application of machine learning models requires both computer science skills and domain knowledge. This presents a barrier for those who want to apply machine learning to real-world settings.

OBJECTIVES

The objective of this study was to construct a competitive predictive model with a minimal amount of human effort to facilitate decisions regarding hospitalization of patients.

METHODS

This study used the electronic health record data from five EDs in a single healthcare system, including an academic urban children's hospital ED, from January 2009 to December 2013. We constructed two machine learning models by using automated machine learning algorithm (autoML) which allows non-experts to use machine learning model: one with data only available at ED triage, the other adding information available one hour into the ED visit. Random forest and logistic regression were employed as bench-marking models. The ratio of the training dataset to the test dataset was 8:2, and the area under the receiver operating characteristic curve (AUC), accuracy, and F1 were calculated to assess the quality of the models.

RESULTS

Of the 9,069 ED visits analyzed, the study population was made up of males (62.7 %), median (IQR) age was 6 (4, 10) years, and public insurance coverage (66.0 %). The majority had an Emergency Severity Index score of 3 (52.9 %). The prevalence of hospitalization was 22.5 %. The AUCs were 0.914 and 0.942. AUCs were 0.831 and 0.886 for random forests, and 0.795 and 0.823 for logistic regression. Among the predictors, an outcome at a prior visit, ESI level, time to first medication, and time to triage were the most important features for the prediction of the need for hospitalization.

CONCLUSIONS

In comparison with the conventional approaches, the use of autoML improved the predictive ability for the need for hospitalization. The findings can optimize ED management, hospital-level resource utilization and improve quality. Furthermore, this approach can support the design of a more effective patient ED flow for pediatric asthma care.

Concurrent Types of Intracranial Hemorrhage are Associated with a Higher Mortality Rate in Adult Patients with Traumatic Subarachnoid Hemorrhage: A Cross-Sectional Retrospective Study

Traumatic subarachnoid hemorrhage (SAH) is the second most frequent intracranial hemorrhage and a common radiologic finding in computed tomography. This study aimed to estimate the risk of mortality in adult trauma patients with traumatic SAH concurrent with other types of intracranial hemorrhage, such as subdural hematoma (SDH), epidural hematoma (EDH), and intracerebral hemorrhage (ICH), compared to the risk in patients with isolated traumatic SAH. We searched our hospital’s trauma database from 1 January, 2009 to 31 December, 2018 to identify hospitalized adult patients ≥20 years old who presented with a trauma abbreviated injury scale (AIS) of ≥3 in the head region. Polytrauma patients with an AIS of ≥3 in any other region of the body were excluded. A total of 1856 patients who had SAH were allocated into four exclusive groups: (Group I) isolated traumatic SAH, n = 788; (Group II) SAH and one diagnosis, n = 509; (Group III) SAH and two diagnoses, n = 493; and (Group IV) SAH and three diagnoses, n = 66. One, two, and three diagnoses indicated occurrences of one, two, or three other types of intracranial hemorrhage (SDH, EDH, or ICH). The adjusted odds ratio with a 95% confidence interval (CI) of the level of mortality was calculated with logistic regression, controlling for sex, age, and pre-existing comorbidities. Patients with isolated traumatic SAH had a lower rate of mortality (1.8%) compared to the other three groups (Group II: 7.9%, Group III: 12.4%, and Group IV: 27.3%, all p < 0.001). When controlling for sex, age, and pre-existing comorbidities, we found that Group II, Group III, and Group IV patients had a 4.0 (95% CI 2.4–6.5), 8.9 (95% CI 4.8–16.5), and 21.1 (95% CI 9.4–47.7) times higher adjusted odds ratio for mortality, respectively, than the patients with isolated traumatic SAH. In this study, we demonstrated that compared to patients with isolated traumatic SAH, traumatic SAH patients with concurrent types of intracranial hemorrhage have a higher adjusted odds ratio for mortality.

An ensemble learning method for asthma control level detection with leveraging medical knowledge-based classifier and supervised learning

Approximately 300 million people are afflicted with asthma around the world, with the estimated death rate of 250,000 cases, indicating the significance of this disease. If not treated, it can turn into a serious public health problem. The best method to treat asthma is to control it. Physicians recommend continuous monitoring on asthma symptoms and offering treatment preventive plans based on the patient's control level. Therefore, successful detection of the disease control level plays a critical role in presenting treatment plans. In view of this objective, we collected the data of 96 asthma patients within a 9-month period from a specialized hospital for pulmonary diseases in Tehran. A new ensemble learning algorithm with combining physicians' knowledge in the form of a rule-based classifier and supervised learning algorithms is proposed to detect asthma control level in a multivariate dataset with multiclass response variable. The model outcome resulting from the balancing operations and feature selection on data yielded the accuracy of 91.66%. Our proposed model combines medical knowledge with machine learning algorithms to classify asthma control level more accurately. This model can be applied in electronic self-care systems to support the real-time decision and personalized warnings on possible deterioration of asthma control level. Such tools can centralize asthma treatment from the current reactive care models into a preventive approach in which the physician's therapeutic actions would be based on control level.

Accurate and rapid screening model for potential diabetes mellitus

Background

Prediction or early diagnosis of diabetes is crucial for populations with high risk of diabetes.

Methods

In this study, we assessed the ability of five popular classifiers (J48, AdaboostM1, SMO, Bayes Net, and Naïve Bayes) to identify individuals with diabetes based on nine non-invasive and easily obtained clinical features, including age, gender, body mass index (BMI), hypertension, history of cardiovascular disease or stroke, family history of diabetes, physical activity, work stress, and salty food preference. A total of 4205 data entries were obtained from annual physical examination reports for adults in the Shengjing Hospital of China Medical University during January–April 2017. Weka data mining software was used to identify the best algorithm for diabetes classification.

Results

The results indicate that decision tree classifier J48 has the best performance (accuracy = 0.9503, precision = 0.950, recall = 0.950, F-measure = 0.948, and AUC = 0.964). The decision tree structure shows that age is the most significant feature, followed by family history of diabetes, work stress, BMI, salty food preference, physical activity, hypertension, gender, and history of cardiovascular disease or stroke.

Conclusions

Our study shows that decision tree analyses can be applied to screen individuals for early diabetes risk without the need for invasive tests. This procedure will be particularly useful in developing regions with high epidemiological risk and poor socioeconomic status, and enable clinical practitioners to rapidly screen patients for increased risk of diabetes. The key features in the tree structure could further facilitate diabetes prevention through targeted community interventions, which can potentially improve early diabetes diagnosis and reduce burdens on the healthcare system.

Provider Prediction of Disposition for Children With an Acute Exacerbation of Asthma Presenting to the Pediatric Emergency Department

OBJECTIVE

The aim of this study was to evaluate the accuracy of the initial impression of emergency department providers on the disposition of children with asthma exacerbation.

METHODS

We conducted a prospective survey of physicians and other providers in the emergency department of a children's hospital and parents of children presenting with asthma exacerbation. The treating provider completed a survey after finishing the examination and immediately upon exiting the patient's room. Providers predicted the disposition of the child. Additionally, the providers indicated the likelihood of admission using several 10-cm visual analog scales (VASs). Physician accuracy was calculated, and logistic regression models and receiver operator characteristic curves were generated.

RESULTS

Complete data were available for 177 subjects. Medical doctors/nurse practitioners made correct predictions in 129 (79.6%; 95% confidence interval [CI], 73.4-85.8) of 162 encounters. Respiratory therapists were correct in 69 (67.6%; 95% CI, 58.6%-76.7%) of 102 encounters, and parents were correct in 116 (67.4%; 95% CI, 60.4%-74.4%) of 172 encounters. Logistic regression with disposition as the dependent variable revealed that provider VAS for likelihood of admission (odds ratio, 23.717; 95% CI, 9.298-60.495) was associated with admission. A receiver operator characteristic curve generated for actual disposition versus "likelihood of admission" VAS had an area under the curve of 0.856 (95% CI, 0.793-0.919). For admission, a VAS of greater than 7 was 89.9% specific, greater than 7.6 was 92.9% specific, and greater than 8.6 was 96% specific.

CONCLUSIONS

Emergency department providers correctly predicted disposition 80% of the time. Providers were particularly likely to correctly predict admission. A VAS score of 7 or greater is nearly 90% specific for admission, with specificity increasing at higher values.

Machine Learning-Based Prediction of Clinical Outcomes for Children During Emergency Department Triage

IMPORTANCE

While machine learning approaches may enhance prediction ability, little is known about their utility in emergency department (ED) triage.

OBJECTIVES

To examine the performance of machine learning approaches to predict clinical outcomes and disposition in children in the ED and to compare their performance with conventional triage approaches.

DESIGN, SETTING, AND PARTICIPANTS

Prognostic study of ED data from the National Hospital Ambulatory Medical Care Survey from January 1, 2007, through December 31, 2015. A nationally representative sample of 52 037 children aged 18 years or younger who presented to the ED were included. Data analysis was performed in August 2018.

MAIN OUTCOMES AND MEASURES

The outcomes were critical care (admission to an intensive care unit and/or in-hospital death) and hospitalization (direct hospital admission or transfer). In the training set (70% random sample), using routinely available triage data as predictors (eg, demographic characteristics and vital signs), we derived 4 machine learning-based models: lasso regression, random forest, gradient-boosted decision tree, and deep neural network. In the test set (the remaining 30% of the sample), we measured the models' prediction performance by computing C statistics, prospective prediction results, and decision curves. These machine learning models were built for each outcome and compared with the reference model using the conventional triage classification information.

RESULTS

Of 52 037 eligible ED visits by children (median [interquartile range] age, 6 [2-14] years; 24 929 [48.0%] female), 163 (0.3%) had the critical care outcome and 2352 (4.5%) had the hospitalization outcome. For the critical care prediction, all machine learning approaches had higher discriminative ability compared with the reference model, although the difference was not statistically significant (eg, C statistics of 0.85 [95% CI, 0.78-0.92] for the deep neural network vs 0.78 [95% CI, 0.71-0.85] for the reference; P = .16), and lower number of undertriaged critically ill children in the conventional triage levels 3 to 5 (urgent to nonurgent). For the hospitalization prediction, all machine learning approaches had significantly higher discrimination ability (eg, C statistic, 0.80 [95% CI, 0.78-0.81] for the deep neural network vs 0.73 [95% CI, 0.71-0.75] for the reference; P < .001) and fewer overtriaged children who did not require inpatient management in the conventional triage levels 1 to 3 (immediate to urgent). The decision curve analysis demonstrated a greater net benefit of machine learning models over ranges of clinical thresholds.

CONCLUSIONS AND RELEVANCE

Machine learning-based triage had better discrimination ability to predict clinical outcomes and disposition, with reduction in undertriaging critically ill children and overtriaging children who are less ill.

Machine learning approaches for predicting disposition of asthma and COPD exacerbations in the ED

OBJECTIVE

The prediction of emergency department (ED) disposition at triage remains challenging. Machine learning approaches may enhance prediction. We compared the performance of several machine learning approaches for predicting two clinical outcomes (critical care and hospitalization) among ED patients with asthma or COPD exacerbation.

METHODS

Using the 2007-2015 National Hospital and Ambulatory Medical Care Survey (NHAMCS) ED data, we identified adults with asthma or COPD exacerbation. In the training set (70% random sample), using routinely-available triage data as predictors (e.g., demographics, arrival mode, vital signs, chief complaint, comorbidities), we derived four machine learning-based models: Lasso regression, random forest, boosting, and deep neural network. In the test set (the remaining 30% of sample), we compared their prediction ability against traditional logistic regression with Emergency Severity Index (ESI, reference model).

RESULTS

Of 3206 eligible ED visits, corresponding to weighted estimates of 13.9 million visits, 4% had critical care outcome and 26% had hospitalization outcome. For the critical care prediction, the best performing approach- boosting - achieved the highest discriminative ability (C-statistics 0.80 vs. 0.68), reclassification improvement (net reclassification improvement [NRI] 53%, P = 0.002), and sensitivity (0.79 vs. 0.53) over the reference model. For the hospitalization prediction, random forest provided the highest discriminative ability (C-statistics 0.83 vs. 0.64) reclassification improvement (NRI 92%, P < 0.001), and sensitivity (0.75 vs. 0.33). Results were generally consistent across the asthma and COPD subgroups.

CONCLUSIONS

Based on nationally-representative ED data, machine learning approaches improved the ability to predict disposition of patients with asthma or COPD exacerbation.

Identification of Pancreatic Injury in Patients with Elevated Amylase or Lipase Level Using a Decision Tree Classifier: A Cross-Sectional Retrospective Analysis in a Level I Trauma Center

Background: In trauma patients, pancreatic injury is rare; however, if undiagnosed, it is associated with high morbidity and mortality rates. Few predictive models are available for the identification of pancreatic injury in trauma patients with elevated serum pancreatic enzymes. In this study, we aimed to construct a model for predicting pancreatic injury using a decision tree (DT) algorithm, along with data obtained from a population-based trauma registry in a Level I trauma center. Methods: A total of 991 patients with elevated serum levels of amylase (>137 U/L) or lipase (>51 U/L), including 46 patients with pancreatic injury and 865 without pancreatic injury between January 2009 and December 2016, were allocated in a ratio of 7:3 to training (n = 642) or test (n = 269) sets. Using the data on patient and injury characteristics as well as laboratory data, the DT algorithm with Classification and Regression Tree (CART) analysis was performed based on the Gini impurity index, using the rpart function in the rpart package in R. Results: Among the trauma patients with elevated amylase or lipase levels, three groups of patients were identified as having a high risk of pancreatic injury, using the DT model. These included (1) 69% of the patients with lipase level ≥306 U/L; (2) 79% of the patients with lipase level between 154 U/L and 305 U/L and shock index (SI) ≥ 0.72; and (3) 80% of the patients with lipase level <154 U/L with abdomen injury, glucose level <158 mg/dL, amylase level <90 U/L, and neutrophil percentage ≥76%; they had all sustained pancreatic injury. With all variables in the model, the DT achieved an accuracy of 97.9% (sensitivity of 91.4% and specificity of 98.3%) for the training set. In the test set, the DT achieved an accuracy of 93.3%, sensitivity of 72.7%, and specificity of 94.2%. Conclusions: We established a DT model using lipase, SI, and additional conditions (injury to the abdomen, glucose level <158 mg/dL, amylase level <90 U/L, and neutrophils ≥76%) as important nodes to predict three groups of patients with a high risk of pancreatic injury. The proposed decision-making algorithm may help in identifying pancreatic injury among trauma patients with elevated serum amylase or lipase levels.

Derivation and validation of different machine-learning models in mortality prediction of trauma in motorcycle riders: a cross-sectional retrospective study in southern Taiwan

OBJECTIVES

This study aimed to build and test the models of machine learning (ML) to predict the mortality of hospitalised motorcycle riders.

SETTING

The study was conducted in a level-1 trauma centre in southern Taiwan.

PARTICIPANTS

Motorcycle riders who were hospitalised between January 2009 and December 2015 were classified into a training set (n=6306) and test set (n=946). Using the demographic information, injury characteristics and laboratory data of patients, logistic regression (LR), support vector machine (SVM) and decision tree (DT) analyses were performed to determine the mortality of individual motorcycle riders, under different conditions, using all samples or reduced samples, as well as all variables or selected features in the algorithm.

PRIMARY AND SECONDARY OUTCOME MEASURES

The predictive performance of the model was evaluated based on accuracy, sensitivity, specificity and geometric mean, and an analysis of the area under the receiver operating characteristic curves of the two different models was carried out.

RESULTS

In the training set, both LR and SVM had a significantly higher area under the receiver operating characteristic curve (AUC) than DT. No significant difference was observed in the AUC of LR and SVM, regardless of whether all samples or reduced samples and whether all variables or selected features were used. In the test set, the performance of the SVM model for all samples with selected features was better than that of all other models, with an accuracy of 98.73%, sensitivity of 86.96%, specificity of 99.02%, geometric mean of 92.79% and AUC of 0.9517, in mortality prediction.

CONCLUSION

ML can provide a feasible level of accuracy in predicting the mortality of motorcycle riders. Integration of the ML model, particularly the SVM algorithm in the trauma system, may help identify high-risk patients and, therefore, guide appropriate interventions by the clinical staff.

Prediction of Mortality in Patients with Isolated Traumatic Subarachnoid Hemorrhage Using a Decision Tree Classifier: A Retrospective Analysis Based on a Trauma Registry System

Background: In contrast to patients with traumatic subarachnoid hemorrhage (tSAH) in the presence of other types of intracranial hemorrhage, the prognosis of patients with isolated tSAH is good. The incidence of mortality in these patients ranges from 0–2.5%. However, few data or predictive models are available for the identification of patients with a high mortality risk. In this study, we aimed to construct a model for mortality prediction using a decision tree (DT) algorithm, along with data obtained from a population-based trauma registry, in a Level 1 trauma center. Methods: Five hundred and forty-five patients with isolated tSAH, including 533 patients who survived and 12 who died, between January 2009 and December 2016, were allocated to training (n = 377) or test (n = 168) sets. Using the data on demographics and injury characteristics, as well as laboratory data of the patients, classification and regression tree (CART) analysis was performed based on the Gini impurity index, using the rpart function in the rpart package in R. Results: In this established DT model, three nodes (head Abbreviated Injury Scale (AIS) score ≤4, creatinine (Cr) <1.4 mg/dL, and age <76 years) were identified as important determinative variables in the prediction of mortality. Of the patients with isolated tSAH, 60% of those with a head AIS >4 died, as did the 57% of those with an AIS score ≤4, but Cr ≥1.4 and age ≥76 years. All patients who did not meet the above-mentioned criteria survived. With all the variables in the model, the DT achieved an accuracy of 97.9% (sensitivity of 90.9% and specificity of 98.1%) and 97.7% (sensitivity of 100% and specificity of 97.7%), for the training set and test set, respectively. Conclusions: The study established a DT model with three nodes (head AIS score ≤4, Cr <1.4, and age <76 years) to predict fatal outcomes in patients with isolated tSAH. The proposed decision-making algorithm may help identify patients with a high risk of mortality.

The role of human rhinovirus (HRV) species on asthma exacerbation severity in children and adolescents

OBJECTIVE

It is recognized that human rhinovirus (HRV) infection is an important factor in asthma exacerbations requiring hospitalization in children. However, previous studies have disagreed on the differential impact of various HRV species. We sought to assess the impact of HRV species on the severity of asthma exacerbations in children and adolescents. We also examined whether the effect of HRV species on severity was modified by age and gender.

METHODS

Virus strain was determined for 113 children with HRV detectable at the time of admission for asthma exacerbation. Patient characteristics were collected on admission and exacerbation severity was scored using several validated scales.

RESULTS

HRV species by itself was not associated with moderate/severe vs. mild exacerbations. Boys with HRV-C infections were more likely (OR: 3.7, 95% CI: 1.2-13.4) to have a moderate/severe exacerbation than girls with HRV-C (p = 0.04 for interaction term). Higher odds were observed in younger boys (3 years old: OR: 9.1, 95% CI: 1.8-47.1 vs 5 years old: OR: 3.3, 95% CI: 0.9-11.8 vs 7 years old: OR: 1.2, 95% CI: 0.2-6.6). In contrast, children with HRV-C infection and sensitized to pollen during the pollen season were less likely to have moderate/severe exacerbations (p = 0.01 for the interaction term).

CONCLUSION

Acute asthma exacerbations are more likely to be moderate/severe in boys under 5 years of age who had HRV-C infection on admission. The opposite was found in children with sensitization to pollen during pollen season.

A Statistical Classifier to Support Diagnose Meningitis in Less Developed Areas of Brazil

This paper describes the development of statistical classifiers to help diagnose meningococcal meningitis, i.e. the most sever, infectious and deadliest type of this disease. The goal is to find a mechanism able to determine whether a patient has this type of meningitis from a set of symptoms that can be directly observed in the earliest stages of this pathology. Currently, in Brazil, a country that is heavily affected by meningitis, all suspected cases require immediate hospitalization and the beginning of a treatment with invasive tests and medicines. This procedure, therefore, entails expensive treatments unaffordable in less developed regions. For this purpose, we have gathered together a dataset of 22,602 records of suspected meningitis cases from the Brazilian state of Bahia. Seven classification techniques have been applied from input data of nine symptoms and other information about the patient such as age, sex and the area they live in, and a 10 cross-fold validation has been performed. Results show that the techniques applied are suitable for diagnosing the meningococcal meningitis. Several indexes, such as precision, recall or ROC area, have been computed to show the accuracy of the models. All of them provide good results, but the best corresponds to the J48 classifier with a precision of 0.942 and a ROC area over 0.95. These results indicate that our model can indeed help lead to a non-invasive and early diagnosis of this pathology. This is especially useful in less developed areas, where the epidemiologic risk is usually high and medical expenses, sometimes, unaffordable.

Comparing predictions made by a prediction model, clinical score, and physicians: pediatric asthma exacerbations in the emergency department

BACKGROUND

Asthma exacerbations are one of the most common medical reasons for children to be brought to the hospital emergency department (ED). Various prediction models have been proposed to support diagnosis of exacerbations and evaluation of their severity.

OBJECTIVES

First, to evaluate prediction models constructed from data using machine learning techniques and to select the best performing model. Second, to compare predictions from the selected model with predictions from the Pediatric Respiratory Assessment Measure (PRAM) score, and predictions made by ED physicians.

DESIGN

A two-phase study conducted in the ED of an academic pediatric hospital. In phase 1 data collected prospectively using paper forms was used to construct and evaluate five prediction models, and the best performing model was selected. In phase 2 data collected prospectively using a mobile system was used to compare the predictions of the selected prediction model with those from PRAM and ED physicians.

MEASUREMENTS

Area under the receiver operating characteristic curve and accuracy in phase 1; accuracy, sensitivity, specificity, positive and negative predictive values in phase 2.

RESULTS

In phase 1 prediction models were derived from a data set of 240 patients and evaluated using 10-fold cross validation. A naive Bayes (NB) model demonstrated the best performance and it was selected for phase 2. Evaluation in phase 2 was conducted on data from 82 patients. Predictions made by the NB model were less accurate than the PRAM score and physicians (accuracy of 70.7%, 73.2% and 78.0% respectively), however, according to McNemar's test it is not possible to conclude that the differences between predictions are statistically significant.

CONCLUSION

Both the PRAM score and the NB model were less accurate than physicians. The NB model can handle incomplete patient data and as such may complement the PRAM score. However, it requires further research to improve its accuracy.

Implementing an integrative multi-agent clinical decision support system with open source software

Clinical decision making is a complex multi-stage process. Decision support can play an important role at each stage of this process. At present, the majority of clinical decision support systems have been focused on supporting only certain stages. In this paper we present the design and implementation of MET3-a prototype multi-agent system providing an integrative decision support that spans over the entire decision making process. The system helps physicians with data collection, diagnosis formulation, treatment planning and finding supporting evidence. MET3 integrates with external hospital information systems via HL7 messages and runs on various computing platforms available at the point of care (e.g., tablet computers, mobile phones). Building MET3 required sophisticated and reliable software technologies. In the past decade the open source software movement has produced mature, stable, industrial strength software systems with a large user base. Therefore, one of the decisions that should be considered before developing or acquiring a decision support system is whether or not one could use open source technologies instead of proprietary ones. We believe MET3 shows that the answer to this question is positive.