3.1.5. Prostate Cancer

Prostate cancer (PCa) is a widespread health concern, affecting 1.3 million men globally in 2018 [90]. Detecting prostate cancer (Pca) in its early stages is vital for effective treatment, and the utilization of biosensors can assist in the early detection of Pca. Furthermore, by utilizing machine learning to analyze the vast amounts of data generated by biosensors, we can achieve highly accurate predictions, ultimately leading to better patient outcomes.

Prostate cancer gene 3 (PCA3) is specifically expressed in the prostate and is strongly associated with prostate cancer. It shows promise as a potential biomarker for detecting prostate cancer. The PCA3 gene has been detected in 95% of prostate cancer samples, making it highly associated with the disease. Even small amounts of PCA3 can indicate a significant likelihood that a patient either has or will develop prostate cancer [91]. As an example, Rodrigues et al. [91] developed a genosensor with carbon-printed electrodes and a layer of a complementary DNA sequence (PCA3 probe). They investigated the ability of electrochemical and optical detection methods, along with machine learning algorithms, to diagnose prostate cancer using images of the genosensors. The study demonstrated that the meta-classifier machine learning algorithms, including SVM and LDA, could accurately classify scanning electron microscopy images with 99.9% accuracy.

Differences in the metabolite components between patient urine and normal urine have been reported, indicating the need for a rapid, easy-to-use, and label-free technique to analyze urine metabolites. Such a technique is crucial for developing on-site urine diagnostic platforms and identifying unknown metabolite biomarkers for cancer detection. In a study conducted by Ling et al. [92], the researchers applied an integrated on-site detection system based on SERS sensor technology and deep learning models to diagnose prostate and pancreatic cancer. The sensor is based on a 3D plasmonic coral nanoarchitecture (3D-PCN) synthesized on a paper substrate, which was integrated with a handheld Raman spectrometer to create an on-site diagnostic platform. Human urine samples are directly absorbed into the paper-based 3D-PCN, and the SERS signals of complicated urine components are obtained without any pretreatment. The RNN and CNN models are employed for the supervised classification of SERS spectra, and the platform achieved high sensitivity and specificity for detecting cancer. The system demonstrates the potential for use as a diagnostic platform in various human biofluid analyses in the future.

The bio-nanochip platform shows promising potential as a versatile and efficient biosensor system for various applications, including medical diagnostics and environmental monitoring. For instance, McRae et al. [93] designed a programmable bio-nanochip (p-BNC) system, a biosensor platform with the capacity for learning. In this system, small quantities of patient samples generate an immunofluorescent signal on agarose bead sensors that are optically extracted and converted to antigen concentrations. This biochip sensor has the potential to detect prostate cancer and ovarian cancer with single-use disposable cartridges. They applied machine learning methods to analyze the dataset.
