Machine Learning Techniques for Prediction of Early Childhood Obesity.

OBJECTIVES: This paper aims to predict childhood obesity after age two, using only data collected prior to the second birthday by a clinical decision support system called CHICA. METHODS: Analyses of six different machine learning methods: RandomTree, RandomForest, J48, ID3, Naïve Bayes, and Bayes trained on CHICA data show that an accurate, sensitive model can be created. RESULTS: Of the methods analyzed, the ID3 model trained on the CHICA dataset proved the best overall performance with accuracy of 85% and sensitivity of 89%. Additionally, the ID3 model had a positive predictive value of 84% and a negative predictive value of 88%. The structure of the tree also gives insight into the strongest predictors of future obesity in children. Many of the strongest predictors seen in the ID3 modeling of the CHICA dataset have been independently validated in the literature as correlated with obesity, thereby supporting the validity of the model. CONCLUSIONS: This study demonstrated that data from a production clinical decision support system can be used to build an accurate machine learning model to predict obesity in children after age two.

Leveraging electronic tablets for general pediatric care: a pilot study.

BACKGROUND: We have previously shown that a scan-able paper based interface linked to a computerized clinical decision support system (CDSS) can effectively screen patients in pediatric waiting rooms and support the physician using evidence based care guidelines at the time of clinical encounter. However, the use of scan-able paper based interface has many inherent limitations including lacking real time communication with the CDSS and being prone to human and system errors. An electronic tablet based user interface can not only overcome these limitations, but may also support advanced functionality for clinical and research use. However, use of such devices for pediatric care is not well studied in clinical settings. OBJECTIVE: In this pilot study, we enhance our pediatric CDSS with an electronic tablet based user interface and evaluate it for usability as well as for changes in patient questionnaire completion rates. METHODS: Child Health Improvement through Computers Leveraging Electronic Tablets or CHICLET is an electronic tablet based user interface. It is developed to augment the existing scan-able paper interface to our CDSS. For the purposes of this study, we deployed CHICLET in one outpatient pediatric clinic. Usability factors for CHICLET were evaluated via caregiver and staff surveys. RESULTS: When compared to the scan-able paper based interface, we observed an 18% increase or 30% relative increase in question completion rates using CHICLET. This difference was statistically significant. Caregivers and staff survey results were positive for using CHICLET in clinical environment. CONCLUSIONS: Electronic tablets are a viable interface for capturing patient self-report in pediatric waiting rooms. We further hypothesize that the use of electronic tablet based interfaces will drive advances in computerized clinical decision support and create opportunities for patient engagement.

Patterns of use and clinical utility of exercise thallium-201 single photon emission-computed tomography in a community hospital.

Single photon emission computed tomographic (SPECT) imaging with thallium-201 was evaluated in 492 consecutive, unselected patients to ascertain the patterns of use and additive value to the exercise electrocardiogram in the diagnosis of myocardial ischemia in a community hospital setting. Myocardial perfusion images were interpreted by a single observer employing visual and confirmatory quantitative analysis. The majority of patients (69%) were studied to evaluate the functional consequences of known coronary artery disease and not primarily to aid in diagnosis. Relationships among exercise electrocardiograms, SPECT information and coronary angiography were analyzed in 303 patients. The sensitivity was 92% and improved to 96% when multivessel disease was present. Overall, the positive predictive value of tomographic imaging in the detection of coronary disease was 94%. These findings demonstrate the changing pattern of use of myocardial perfusion imaging and confirm the value and applicability of this technique in routine clinical practice.

Transient global amnesia associated with bradycardia and temporal lobe spikes.

This paper presents electroencephalographic evidence of bilateral independent temporal lobe spiking during an episode of transient global amnesia. The amnesia occurred during a period of hypotension secondary to sinus bradycardia. The amnesia recurred in absence of bradycardia and hypotension. The temporal lobe spiking leads credence to the thought that transient global amnesia occurs as the result of seizure activity.

Lateral medullary syndrome after neck injury.

Several reported cases of brainstem stroke followed neck manipulation or trauma. We describe a 28-year-old man who had a lateral medullary syndrome after cervical hyperextension in an automobile accident. Angiography revealed complete occlusion of the left vertebral artery.