Electronic Theses and Dissertation

Stres merupakan kondisi psikologis yang ditandai oleh ketegangan mental dan emosional yang dapat memengaruhi keseimbangan individu. Konsultasi dengan tenaga profesional menjadi salah satu cara penanganan stres, namun sering terkendala waktu, biaya, dan keengganan individu untuk mencari bantuan langsung. Oleh karena itu, penelitian ini mengusulkan pengembangan aplikasi mobile berbasis Android bernama S-Check, yang dirancang menggunakan framework Flutter untuk mendeteksi tingkat stres melalui citra wajah. Model yang digunakan adalah EfficientNet-B0 dalam format TensorFlow Lite (TFLite) agar aplikasi dapat beroperasi secara offline. Metode penelitian menerapkan HumanCentered Design (HCD) dengan pengujian menggunakan Black Box Testing dan System Usability Scale (SUS), serta validasi melalui kuesioner Depression Anxiety Stress Scales (DASS-42). Berdasarkan pengujian terhadap 12 responden, diperoleh tingkat kesesuaian antara hasil deteksi aplikasi dan DASS-42 sebesar 33,3%, dengan perbedaan hasil yang dipengaruhi oleh faktor pencahayaan, ekspresi wajah, posisi kamera, dan kondisi emosional saat pengambilan gambar. Aplikasi juga memperoleh skor 79 pada pengujian SUS, yang termasuk dalam kategori “Baik”, sehingga dinilai layak digunakan. Meskipun terdapat perbedaan hasil dengan DASS-42, S-Check tetap efektif sebagai alat bantu deteksi stres dini dan dapat meningkatkan kesadaran pengguna terhadap pentingnya kesehatan mental.

Stress is a psychological condition characterized by mental and emotional tension that can affect an individual's balance. Consultation with mental health professionals is one of the methods to manage stress; however, it often faces obstacles such as time constraints, costs, and individuals’ reluctance to seek direct help. Therefore, this study proposes the development of a mobile application called S-Check, designed using the Flutter framework to detect stress levels through facial images. The model used is EfficientNet-B0 in TensorFlow Lite (TFLite) format to enable the application to operate offline. The research methodology applies the Human-Centered Design (HCD) approach, with testing conducted using Black Box Testing and the System Usability Scale (SUS), as well as validation through the Depression Anxiety Stress Scales (DASS-42) questionnaire. Based on testing with 12 respondents, the level of agreement between the application’s detection results and the DASS-42 was 33.3%, with differences influenced by factors such as lighting conditions, facial expressions, camera positioning, and the respondents’ emotional states during image capture. The application also obtained a score of 79 on the SUS test, which falls into the “Good” category, indicating that it is feasible for use. Despite differences with the DASS-42 results, S-Check remains effective as an early stress detection tool and can increase users’ awareness of the importance of mental health.