A novel self-powered, highly selective acetone sensor based on InP/Pt/chitosan nanowire arrays has been developed for diabetes monitoring. The sensor can detect acetone levels from sub-ppb to over 100,000 ppm at room temperature, covering the range of acetone in exhaled breath from healthy individuals (300–800 ppb) to those at high risk of diabetic ketoacidosis (DKA) (>75 ppm). The sensor integrates a chitosan surface-functional layer and a Pt Schottky contact to enable efficient charge transfer and photovoltaic effects, achieving high sensitivity and selectivity. The sensor was successfully integrated into a handheld breath testing prototype, the Ketowhistle, which can detect acetone concentrations in simulated breath samples. The Ketowhistle demonstrates the potential for non-invasive ketone monitoring, particularly for DKA prevention. The sensor's performance was evaluated through controlled experiments and simulations, revealing an oxygen-facilitated two-step charge transfer process for acetone reduction. The sensor exhibits rapid response and recovery times, high sensitivity, and excellent stability, making it suitable for real-time monitoring. The device was tested with simulated breath samples and showed good consistency in distinguishing diabetic from non-diabetic breath, even under high humidity conditions. The study highlights the potential of the InP/Pt/chitosan nanowire sensor for future reliable breath analysis in diabetes diagnosis and management.A novel self-powered, highly selective acetone sensor based on InP/Pt/chitosan nanowire arrays has been developed for diabetes monitoring. The sensor can detect acetone levels from sub-ppb to over 100,000 ppm at room temperature, covering the range of acetone in exhaled breath from healthy individuals (300–800 ppb) to those at high risk of diabetic ketoacidosis (DKA) (>75 ppm). The sensor integrates a chitosan surface-functional layer and a Pt Schottky contact to enable efficient charge transfer and photovoltaic effects, achieving high sensitivity and selectivity. The sensor was successfully integrated into a handheld breath testing prototype, the Ketowhistle, which can detect acetone concentrations in simulated breath samples. The Ketowhistle demonstrates the potential for non-invasive ketone monitoring, particularly for DKA prevention. The sensor's performance was evaluated through controlled experiments and simulations, revealing an oxygen-facilitated two-step charge transfer process for acetone reduction. The sensor exhibits rapid response and recovery times, high sensitivity, and excellent stability, making it suitable for real-time monitoring. The device was tested with simulated breath samples and showed good consistency in distinguishing diabetic from non-diabetic breath, even under high humidity conditions. The study highlights the potential of the InP/Pt/chitosan nanowire sensor for future reliable breath analysis in diabetes diagnosis and management.