Researchers at the University of Jordan have developed REWAA, a bilingual mobile application combining mood tracking, computer vision facial analysis, and therapist communication to support digital mental health management.
Introduction to REWAA App REWAA is an innovative bilingual Android application created by computer engineering researchers at the University of Jordan. Designed for digital mental health support, it offers multimodal emotional tracking and direct interaction with licensed therapists, catering to both Arabic- and English-speaking users. The app addresses a critical global health challenge: depressive disorders, particularly highlighting the significant gap in accessible and culturally appropriate digital mental health resources for Arabic-speaking populations. Development and Architecture The development of REWAA involved a comprehensive co-design phase, incorporating insights from educational specialists, family caregivers, and clinical software advisors through moderated focus groups. This collaborative approach helped establish core user requirements, localized cultural expectations, and visual accessibility needs. The resulting architecture features distinct patient and therapist account portals. The application client was built using Android Studio, with Google Firebase serving as the robust backend, managing user authentication, real-time data synchronization, cloud storage, and push notifications across devices. Bilingual Hybrid Mood Heuristic Protocol A central component of REWAA's tracking framework is its unique Bilingual Hybrid Mood Heuristic protocol. This experimental mechanism integrates both subjective psychological reporting and objective automated computer vision metrics. Users complete the Patient Health Questionnaire-9 for self-reported depressive symptoms. Simultaneously, an adapted computer vision module (utilizing Google Machine Learning Kit and BlazeFace CNN model) analyzes facial landmark displacements. The heuristic framework then combines these two inputs with a 50/50 weighting, yielding a comprehensive mood tracking value between 0 and 27 to guide users towards appropriate resources or professional consultation. Biometric Tracking Engine Details The underlying biometric tracking engine in REWAA evaluates five specific normalized facial landmark features continuously. These parameters include smile deficit (mouth commissure distance), eye fatigue (eyelid aperture and blink rate), head posture (downward flexion and forward tilt), facial stability (micro-tremor and psychomotor retardation shifts), and a normalized face size indicator. This sophisticated system is designed to monitor subtle behavioral shifts without claiming clinical diagnostic validity. To ensure data accuracy, a mandatory two-minute observational sampling loop processes between 3,600 and 7,200 video frames, automatically restarting if lighting conditions are poor or head movements disrupt data collection. REWAA Bot and Crisis Support For immediate emotional guidance, REWAA incorporates a conversational AI tool, the 'REWAA bot,' powered by the GPT-4o-mini large language model, offering dedicated bilingual capabilities in Arabic and English. User interactions are processed through a multi-stage pipeline that detects language, strips metadata, and maintains conversational continuity with a rolling memory of the last 20 messages. Crucially, an upstream rule-based crisis intercept layer constantly scans input for high-risk terms or self-harm indicators, instantly bypassing the generative model to provide emergency contacts and crisis resources, though the clinical efficacy of this intercept was not independently validated in the study. Self-Guided Wellness Modules Beyond algorithmic tracking and AI-powered conversation, REWAA provides a rich array of self-guided psychoeducational and behavioral activation modules aimed at fostering constructive daily routines. These resources include structured video courses on depression, motivational messaging feeds, guided physical and breathing exercises, and soothing environmental audio tracks. To encourage engagement, a gamified 'mental health tree' visually expands as users consistently complete wellness activities and log positive mood trajectories. For acute distress, an emergency center offers quick communication shortcuts to designated family members, caregivers, or national emergency services. Therapist Workflow and Telehealth Therapist accounts within REWAA are protected by stringent administrative credentialing protocols, ensuring access is limited to certified healthcare providers. Practitioners must submit official practice licenses and institutional certifications for manual verification. Once approved, therapists gain access to a dashboard allowing them to review patient appointment requests, manage clinical calendars, record private clinical notes separate from patient records, and conduct synchronous virtual consultations. The telehealth features include integrated live video session rooms with timers, recording toggles, and archived session histories. Patient privacy is paramount, with all communication encrypted and local data cached securely on user devices. System Performance and Reliability System performance and cross-platform reliability were rigorously verified across a diverse range of seven commercial Android smartphones (Android versions 9-13), including popular brands like Samsung, Google Pixel, and Xiaomi. Automated technical benchmarks revealed stable chatbot response latencies, typically ranging between 1.5 and 3.0 seconds, alongside impressive language identification and response accuracy rates exceeding 98%. The core application services, including the Mood Heuristic test, live therapist communication, patient profile management, and interactive support groups, operated reliably across both modern mid-range and flagship hardware. User Evaluation and Feedback A 30-day evaluation trial involving 100 participants from University of Jordan medical clinics and online platforms was conducted to assess practical software usability, perceived usefulness, and interface acceptance. The cohort, comprising 57 females and 43 males with varied demographics and depression diagnoses, provided highly favorable feedback. 85% rated ease of use as very or fairly easy, 70% found the GUI comfortable, and 60% found the REWAA bot provided practical guidance. Significant improvements were noted in personal emotional awareness (55%), trust in data privacy (60%), and perceived benefit alongside psychotherapy (65%), with 60% recommending the app to others. Limitations and Future Development The authors acknowledged several methodological limitations, emphasizing that the study focused on user acceptance rather than clinical diagnostic accuracy or therapeutic effectiveness. Constraints included a modest sample size, a limited 30-day testing window lacking longitudinal tracking, and no formal comparison against gold-standard clinical diagnostic interviews. Technical challenges persist in speech and facial analysis due to lighting, camera quality, and regional Arabic dialect variations. Future initiatives aim to enhance dialectical Arabic language modeling, expand language support, and conduct formal randomized controlled trials with psychiatric institutions to establish clinical reliability and therapeutic validity.
Introduction to REWAA App
REWAA is an innovative bilingual Android application created by computer engineering researchers at the University of Jordan. Designed for digital mental health support, it offers multimodal emotional tracking and direct interaction with licensed therapists, catering to both Arabic- and English-speaking users. The app addresses a critical global health challenge: depressive disorders, particularly highlighting the significant gap in accessible and culturally appropriate digital mental health resources for Arabic-speaking populations.
Development and Architecture
The development of REWAA involved a comprehensive co-design phase, incorporating insights from educational specialists, family caregivers, and clinical software advisors through moderated focus groups. This collaborative approach helped establish core user requirements, localized cultural expectations, and visual accessibility needs. The resulting architecture features distinct patient and therapist account portals. The application client was built using Android Studio, with Google Firebase serving as the robust backend, managing user authentication, real-time data synchronization, cloud storage, and push notifications across devices.
Bilingual Hybrid Mood Heuristic Protocol
A central component of REWAA's tracking framework is its unique Bilingual Hybrid Mood Heuristic protocol. This experimental mechanism integrates both subjective psychological reporting and objective automated computer vision metrics. Users complete the Patient Health Questionnaire-9 for self-reported depressive symptoms. Simultaneously, an adapted computer vision module (utilizing Google Machine Learning Kit and BlazeFace CNN model) analyzes facial landmark displacements. The heuristic framework then combines these two inputs with a 50/50 weighting, yielding a comprehensive mood tracking value between 0 and 27 to guide users towards appropriate resources or professional consultation.
Biometric Tracking Engine Details
The underlying biometric tracking engine in REWAA evaluates five specific normalized facial landmark features continuously. These parameters include smile deficit (mouth commissure distance), eye fatigue (eyelid aperture and blink rate), head posture (downward flexion and forward tilt), facial stability (micro-tremor and psychomotor retardation shifts), and a normalized face size indicator. This sophisticated system is designed to monitor subtle behavioral shifts without claiming clinical diagnostic validity. To ensure data accuracy, a mandatory two-minute observational sampling loop processes between 3,600 and 7,200 video frames, automatically restarting if lighting conditions are poor or head movements disrupt data collection.
REWAA Bot and Crisis Support
For immediate emotional guidance, REWAA incorporates a conversational AI tool, the 'REWAA bot,' powered by the GPT-4o-mini large language model, offering dedicated bilingual capabilities in Arabic and English. User interactions are processed through a multi-stage pipeline that detects language, strips metadata, and maintains conversational continuity with a rolling memory of the last 20 messages. Crucially, an upstream rule-based crisis intercept layer constantly scans input for high-risk terms or self-harm indicators, instantly bypassing the generative model to provide emergency contacts and crisis resources, though the clinical efficacy of this intercept was not independently validated in the study.
Self-Guided Wellness Modules
Beyond algorithmic tracking and AI-powered conversation, REWAA provides a rich array of self-guided psychoeducational and behavioral activation modules aimed at fostering constructive daily routines. These resources include structured video courses on depression, motivational messaging feeds, guided physical and breathing exercises, and soothing environmental audio tracks. To encourage engagement, a gamified 'mental health tree' visually expands as users consistently complete wellness activities and log positive mood trajectories. For acute distress, an emergency center offers quick communication shortcuts to designated family members, caregivers, or national emergency services.
Therapist Workflow and Telehealth
Therapist accounts within REWAA are protected by stringent administrative credentialing protocols, ensuring access is limited to certified healthcare providers. Practitioners must submit official practice licenses and institutional certifications for manual verification. Once approved, therapists gain access to a dashboard allowing them to review patient appointment requests, manage clinical calendars, record private clinical notes separate from patient records, and conduct synchronous virtual consultations. The telehealth features include integrated live video session rooms with timers, recording toggles, and archived session histories. Patient privacy is paramount, with all communication encrypted and local data cached securely on user devices.
System Performance and Reliability
System performance and cross-platform reliability were rigorously verified across a diverse range of seven commercial Android smartphones (Android versions 9-13), including popular brands like Samsung, Google Pixel, and Xiaomi. Automated technical benchmarks revealed stable chatbot response latencies, typically ranging between 1.5 and 3.0 seconds, alongside impressive language identification and response accuracy rates exceeding 98%. The core application services, including the Mood Heuristic test, live therapist communication, patient profile management, and interactive support groups, operated reliably across both modern mid-range and flagship hardware.
User Evaluation and Feedback
A 30-day evaluation trial involving 100 participants from University of Jordan medical clinics and online platforms was conducted to assess practical software usability, perceived usefulness, and interface acceptance. The cohort, comprising 57 females and 43 males with varied demographics and depression diagnoses, provided highly favorable feedback. 85% rated ease of use as very or fairly easy, 70% found the GUI comfortable, and 60% found the REWAA bot provided practical guidance. Significant improvements were noted in personal emotional awareness (55%), trust in data privacy (60%), and perceived benefit alongside psychotherapy (65%), with 60% recommending the app to others.
Limitations and Future Development
The authors acknowledged several methodological limitations, emphasizing that the study focused on user acceptance rather than clinical diagnostic accuracy or therapeutic effectiveness. Constraints included a modest sample size, a limited 30-day testing window lacking longitudinal tracking, and no formal comparison against gold-standard clinical diagnostic interviews. Technical challenges persist in speech and facial analysis due to lighting, camera quality, and regional Arabic dialect variations. Future initiatives aim to enhance dialectical Arabic language modeling, expand language support, and conduct formal randomized controlled trials with psychiatric institutions to establish clinical reliability and therapeutic validity.
