Artificially Intelligent Olfaction for Fast and Noninvasive Diagnosis of Bladder Cancer from Urine

Abstract

Globally, bladder cancer (BLC) is one of the most common cancers and has a high recurrence and mortality rate. Current clinical diagnostic approaches are either invasive or inaccurate. Here, we report on a cost-efficient, artificially intelligent chemiresistive sensor array made of polyaniline (PANI) derivatives that can noninvasively diagnose BLC at an early stage and maintain postoperative surveillance through ″smelling″ clinical urine samples at room temperature. In clinical trials, 18 healthy controls and 76 BLC patients (60 and 16 at early and advanced stages, respectively) are assessed by the artificial olfactory system. With the assistance of a support vector machine (SVM), very high sensitivity and accuracy from healthy controls are achieved, exceeding those obtained by the current techniques in practice. In addition, the recurrences of both early and advanced stages are diagnosed well, with the effect of confounding factors on the performance of the artificial olfactory system found to have a negligible influence on the diagnostic performance. Overall, this study contributes a novel, noninvasive, easy-to-use, inexpensive, real-time, accurate method for urine disease diagnosis, which can be useful for personalized care/diagnosis and postoperative surveillance, resulting in saving more lives.

Modular Assembly of MXene Frameworks for Noninvasive Disease Diagnosis via Urinary Volatiles

“…Urinary samples are rich in chemical markers, including metabolites, proteins, and different VOCs. , Detecting urinary VOCs has great potential to diagnose diseases with a range of VOCs markers (acids, alcohols, ketones, aldehydes, amines, N-heterocycles, O-heterocycles, sulfur compounds, and hydrocarbons) in urine. − One single sensor has extreme difficulty in recognizing multiple VOCs, but an e-nose can detect multiple VOCs via cross-reactivity of the sensor array. Therefore, a sensor array (i.e., e-nose) with highly discriminative recognition is crucial for real applications.…”

Polyaniline-Based Biological and Chemical Sensors: Sensing Mechanism, Configuration Design, and Perspective

“…152−154 For example, Wu et al constructed an AI E-nose with different doped PANI thin films with/without plasma treatment, which displayed a unique and reproducible cross-reactive sensing performance for 11 VOCs associated with bladder cancer (BLC) (Figure 6a). 155 Supportive vector machine algorithms were applied to enable the sensor array to diagnose whether a person has BLC by urine detection. Very high sensitivity (100%), specificity (83.33%), and accuracy (96.67%) were achieved in the diagnosis of clinical urinary samples collected from 76 BLC patients.…”

“…In order to achieve intelligent bio/chemical sensors, a series of data analysis strategies (e.g., principal component analysis (PCA), linear discrimination analysis (LDA)) are required to process the massive amount of data. With the assistance of integrated sensor arrays, it can be expected to achieve high selectivity and identify complex samples via extracting multi-modal information from the analytes and processing the signals in combination with artificial intelligence (AI) algorithms. − For example, Wu et al constructed an AI E-nose with different doped PANI thin films with/without plasma treatment, which displayed a unique and reproducible cross-reactive sensing performance for 11 VOCs associated with bladder cancer (BLC) (Figure a) . Supportive vector machine algorithms were applied to enable the sensor array to diagnose whether a person has BLC by urine detection.…”

“…(a) Combining sensor arrays with artificial intelligence algorithms to create volatile urine sensors that can be used for bladder cancer diagnosis. Reproduced with permission from ref . Copyright 2022 American Chemical Society.…”

Advances in Noble Metal-Decorated Metal Oxide Nanomaterials for Chemiresistive Gas Sensors: Overview

“…With the rapid development of Internet of Things (IoTs), highly sensitive and selective gas sensors with remarkably low limit of detection (LOD), fast response/recovery speed, and excellent long-term stability and reversibility are in ever increasing demand for smart cities, smart plants, and even smart healthcare [1][2][3][4]. On the one hand, ultra-sensitive gas sensors can monitor even trace level hazardous, toxic, or explosive gases like volatile organic compounds (VOCs), hydrogen sulfide (H 2 S), ammonia (NH 3 ), formaldehyde (HCHO), nitrogen dioxide (NO 2 ), methane (CH 4 ), and hydrogen (H 2 ), protecting human health from environmental pollutants or leakage accidents [5][6][7].…”