Potentiometric multisensor system as a possible simple tool for non-invasive prostate cancer diagnostics through urine analysis

• Published in: Sensors and actuators. B, Chemical Vol. 289; pp. 42 - 47
• Main Authors: Solovieva, Svetlana, Karnaukh, Mikhail, Panchuk, Vitaly, Andreev, Evgeny, Kartsova, Liudmila, Bessonova, Elena, Legin, Andrey, Wang, Ping, Wan, Hao, Jahatspanian, Igor, Kirsanov, Dmitry
• Format: Journal Article
• Language: English
• Published: Lausanne Elsevier B.V 15.06.2019; Elsevier Science Ltd
• Subjects: Chemical fingerprinting; Classification; Diagnostics; Logistic regression; Organic chemistry; Potentiometric multisensor system; Prostate cancer; Regression analysis; Regression models; Urine; Logistic regression; Potentiometric multisensor system; Diagnostics; Prostate cancer; Classification
• ISSN: 0925-4005; 1873-3077
• DOI: 10.1016/j.snb.2019.03.072

•Potentiometric multisensor system for urine analysis.•Machine learning algorithms for urine samples classification.•Non-invasive prostate cancer detection with high sensitivity and specificity. We report a simple potentiometric multisensor system for distinguishing the urine samples from the patients with diagnosed prostate cancer and from healthy control group. The sensors of the system are sensitive towards variety of cationic and anionic species in urine, as well as to the presence of RedOx pairs. The response of the system represents a complex chemical fingerprint of urine sample that can be related with patient status through multivariate modelling. 89 urine samples were studied (43 from cancer patients confirmed by prostatic puncture biopsy and 46 from healthy control group) and variety of multivariate classification techniques was applied to the potentiometric data. The best results were obtained with logistic regression model yielding 100% sensitivity and 93% specificity in the independent test set of samples.