DOI Services Persistent Identifiers for Scholarly Research
You are viewing the preserved copy. Open the article on the journal's website →

Contemporary Journal of Social Science Review • pp. 1434~1450

PREDICTIVE URINALYSIS USING AI AND MACHINE LEARNING

  • Rabia Alam*1
  • Sohail Masood Bhatti1
  • Ali Ahmad Sheikh2
  • Sundus Sufyan2

Abstract

Usage of artificial intelligence in healthcare system, in recent years, to interpret the data and to predict the disease has increased extensively. AI tools use binary classification system to interpret the data either as “Positive” or “Negative”. Such interpretation when used in urinalysis has limited clinical benefits. The current study is designed to improve the machine learning (ML) models such as Support Vector Machines, Artificial Neural Networks and Random Forest and to evaluate their output to differentially diagnose urinalysis data on 20 different parameters. In current study, urinalysis reports of 800 patients were collected. The record was anonymized to hide identity of the patients. Out of these 800 test reports, 640 (80%) reports were used to train the models while 160 (20%) reports were used to evaluate the efficiency of the trained models. The dataset was pre-processed for feature scaling and removing any missing values. To determine the performance assessment of the AI driven models, diverse array of standardized metrics was applied. The outcomes of this study showed that AI tools have potential for rapid, accurate diagnosis based of urinalysis and can be further use as non-invasive diagnostic tool. It is pertinent to mention that all previously conducted research work were on single parameters of urinary tract related diseases, while the current study is capable of predicting six diseases based on differential diagnosis with 93.29% accuracy which is much better than previously conducted research work. This advancement could lead to improved patient’s disease diagnosis and will facilitate clinical workflows. It is anticipated that future research will focus on expanding the dataset and investigating deeper learning techniques.

Record

Identifier
https://doi-services.com/1102/3003
DOI
10.66857/acad1r51
Published
2026-03-09
Version
1
License
https://creativecommons.org/licenses/by/4.0
Original
https://studentsconsultancyhome.com/contemporary-journal/index.php/cjssr/article/view/3003