File(s) under embargo
until file(s) become available
Predictive Analysis on the Onset of Complete Renal Failure for Patients with Chronic Kidney Disease
thesisposted on 01.05.2021, 00:00 by Piervincenzo Ventrella
Background and objective The Chronic Kidney Disease (CKD) provokes the gradual loss of kidney function, forcing patients in the final stage undergoing permanent kidney replacement therapy. It affects 12-14% of people worldwide and its related care costs represent an important percentage of the total health expenditure. The most effective weapons against the disease are early diagnosis and treatment, which in the majority of the cases can only postpone the onset of complete kidney failure. CKD patients are classified trough the estimated Glomerular Filtration Rate (eGFR) clinical test, which is currently used as assessment scale for planning follow up and management. To promote personalized care and strategic planning of the treatment, more effective assessment tools are needed. Methods To accurately predict the time frame within which a CKD patient will have necessarily to be dialyzed, thus allowing both the patients and the hospital to organize in the most appropriate manner, we developed a computational model based on a supervised machine learning approach. We compare different techniques, regarding both the information extraction and model training phases, in order to understand which approaches are the most effective. The different models to be compared are trained on the data extracted from the Electronic Medical Record of a hospital infrastructure. Results As final model, we propose a set of Extremely Randomized Trees classifiers considering 27 features, including creatinine level, urea, red blood cells count, eGFR trend (which is not even the most important), age and associated comorbidities. In predicting the occurrence of complete renal failure within the next year or later, it obtains a test accuracy of 94%, specificity of 91% and sensitivity of 96%. More and shorter time-frame intervals, up to 6 months of granularity, can be specified without relevantly worsening the model performance. Conclusions The developed computational model provides clinicians with a great support in predicting the patient’s clinical pathway. Nephrologist showed great interest in the approach and its possible applications. The promising results of the model, coupled with the knowledge and experience of the clinicians, can effectively lead to better personalized care and strategic planning of both patient needs and hospital resources.