Statistical Analysis of Blood and Renal Biomarkers in Chemotherapy Patients Through AI-Based Validation

Background and Aims: In recent years, Artificial Intelligence (AI) and machine learning have increasingly influenced medical data analysis. While AI tools like ChatGPT Plus, Grok۳, and DeepSeek have emerged, the use of traditional software such as SPSS, Prism, and Matlab has declined due to AI’s ability to generate more extensive outputs. This study presents a validation framework for AI tools in analyzing statistical data from cancer patients undergoing treatment, and assessing their ability to process biochemical and hematological parameters. Methods: This study examined ۴۵۷ chemotherapy patients. Urea and creatinine levels were categorised as normal or abnormal, with creatinine also classified by gender. Platelet (Plt), hemoglobin (Hb), and white blood cell (WBC) counts were analyzed within their normal and abnormal ranges. Statistical analyses were first conducted using Prism (version ۱۰.۴.۱) and then replicated using AI-based platforms, including ChatGPT Plus, Grok۳, DeepSeek, Poe, Perplexity, Gemini, Monica, and Sciences. However, only ChatGPT Plus, Grok۳, and DeepSeek provided relevant statistical outputs. The results from these three AI tools were compared with Prism to evaluate consistency and reliability. Results: An independent t-test was performed, assuming a Gaussian distribution with equal standard deviation. Comparison of P-values across different tools showed varying agreement with Prism. Results for Prism versus DeepSeek: Plt = ۰.۱۳۴۷, Hb = ۰.۴۷۵۴, WBC = ۰.۰۵۱۲. Prism versus Grok۳: Plt = ۰.۷۴۰۹, Hb = ۰.۶۵۷۳, WBC = ۰.۴۵۸۴, indicating Grok۳’s closer alignment with Prism. ChatGPT Plus demonstrated near-perfect agreement (P-Value ۰.۹۹۹), producing identical results to Prism. Conclusion: The study indicates that ChatGPT Plus, with its high accuracy and results comparable to Prism, can be a suitable alternative. Grok۳ also performed well, but DeepSeek showed weakness in WBC measurement and was less reliable. AI tools are emerging as effective alternatives to statistical software, but their accuracy in specific variables should be independently validated. Keywords: Comparative analysis, AI, Patient Markers, Chemotherapy

کلیدواژه ها:

Comparative analysis ، AI ، Patient Markers ، Chemotherapy