Human-computer interaction (HCI) strategies communicate the human mind and machine intelligence based on different devices and technologies. The majority of HCI strategies assume normal physical conditions that limit accessibility for users with disabilities. Certain products, such as Braille keyboards, work fine for people with specific disabilities. However, a more general HCI strategy that can neglect users’ physical conditions would enhance the accessibility of these tools for disabled persons. Here, we report an HCI strategy that utilises triboelectricity of the human body (TEHB) for HCI. The TEHB can be generated by many parts of the human body, eliminating the obstacles imposed by physical function disabilities. Such an HCI approach has been used for text inputs, graphical inputs, and mimicked mouse functions. With the assistance of deep learning, an accuracy of approximately 98.4 % is achieved for text inputs obtained directly from handwriting. Our findings provide a new approach for HCI and demonstrate the feasibility of multiple interaction modes.