Deep self-supervised machine learning algorithms with a novel feature elimination and selection approaches for blood test-based multi-dimensional health risks classification.

BACKGROUND: Blood test is extensively performed for screening, diagnoses and surveillance purposes. Although it is possible to automatically evaluate the raw blood test data with the advanced deep self-supervised machine learning approaches, it has not been profoundly investigated and implemented yet. RESULTS: This paper proposes deep machine learning algorithms with multi-dimensional adaptive feature elimination, self-feature weighting and novel feature selection approaches. To classify the health risks based on the processed data with the deep layers, four machine learning algorithms having various properties from being utterly model free to gradient driven are modified. CONCLUSIONS: The results show that the proposed deep machine learning algorithms can remove the unnecessary features, assign self-importance weights, selects their most informative ones and classify the health risks automatically from the worst-case low to worst-case high values.

Thorax CT Dose Reduction Based on Patient Features: Effect of Patient Characteristics on Image Quality and Effective Dose.

Computed tomography (CT) radiation dose reduction is vital without compromising image quality. The aim was to determine the effects of patient characteristics on the received radiation dose and image quality in chest CT examinations and to be able to predict dose and image quality prior to scanning. Consecutive 230 patients underwent routine chest CT examinations were included. CT examination and patients input parameters were recorded for each patient. The effect of patients' demographics/anthropometrics on received dose and image quality was investigated by linear regression analysis. All parameters were evaluated using an artificial neural network (ANN). Of all parameters, patient demographics/anthropometrics were found to be 98% effective in calculating dose reduction. Using ANN on 60 new patients was more than 90% accurate for output parameters and 91% for image quality. Patient characteristics have a significant impact on radiation dose and image quality. Dose and image quality can be determined before CT. This will allow setting the most appropriate scanning parameters before the CT scan.