Page 52 - חוברת תזות הנדסה ירוקה 2023

to determine whether the patients are infected or not. In the next step, the infected samples were classified as being bacterial or viral. The final step was to monitor the efficacy of the antibiotic treatment given to those patients suffering from bacterial infections. To monitor the effectiveness of each antibiotic treatment, the WBCs and serum of the patients with bacterial infections were measured from the first day of treatment and for the next few days. We monitored the spectral changes in the WBCs and the serum daily, to check whether the patients were responding to their treatments. After that, the data were analyzed by different methods, to determine the effectiveness of the antibiotic treatments they were given.
Our findings indicate that the etiology infections can be determined as being bacterial or viral within an hour after receiving the blood samples, with a success rate of over 90%, for both accessible and inaccessible infections. The analyses of the medical indices )for accessible infections only, based on the results of the classification of laboratory samples - Gold Standard( show that the the CRP measure is the most effective for determining the type of infection.
The logistic regression )LR( classifier was able to monitor the effectiveness of the antibiotic treatment based on the spectra of the WBCs. The patients’ bacterial or control statuses were predicted daily, and the probability of each sample belonging to the ‘control’ category was also calculated. The results of these analyses show an accuracy of over 89% in monitoring the effectiveness of the antibiotic treatment, allowing us to conclude that the treatments being given to the patients were appropriate and effective.
Keywords: Bacterial and Viral Infections, White Blood Cells, Immune System, Machine Learning, Efficacy of Antibiotic Therapy.
Book Of Abstracts | Class 2022
52