Tele-OTIVA is a mobile phone-based technology that utilizes an AI model to detect precancerous cervical lesions using cervicography data. The model's performance was tested through several stages: evaluating the AI model, having oncology-gynecologists assess cervicography images taken by healthcare workers, and conducting direct tests on 1,611 women aged 20 to 50 at Dr. Mohammad Hoesin Hospital in Palembang and 12 primary health centers. Twelve healthcare workers were trained to use the application, called the TeleOTIVA App. Four mobile phones—OVO, Infinix, Xiaomi Redmi, and Samsung—were used to evaluate the processing speed and quality of VIA images.
The system architecture employs a detailed approach, using Python/PyTorch for cervical image segmentation, detection, and classification through the YoloSegv8 architecture. Golang is responsible for create, read, update, and delete (CRUD) operations, while Flutter is used for mobile app development. A cloud server handles data processing and learning tasks. Docker is employed to manage the server and prevent service overlap, and SQL is used to store data from various system processes, including medical record input, image prediction, treatment, and recommendations from oncology-gynecology subspecialists. Product development will focus on enhancing patient databases, improving service detection systems, optimizing UI/UX interfaces, and strengthening data and information security measures (see Figure 1).