AI-assisted Pediatric Telemedicine for Remote Growth Monitoring

Thesis Abstract

The study addresses the persistent gap in timely growth assessment for children in underserved regions by leveraging AI-driven telemedicine to enable remote monitoring of pediatric growth trajectories and early identification of growth faltering. The aim is to evaluate the effectiveness, feasibility, and acceptance of an AI-assisted telemedicine platform that integrates routinely collected anthropometric data, parent-reomote reporting, and clinic-based measurements to support growth monitoring and clinical decision-making. Specific objectives include (1) developing and validating an AI model that predicts short- and medium-term growth trajectories using longitudinal anthropometric data, nutrition, and activity indicators; (2) evaluating the accuracy and reliability of remote measurements compared with in-clinic assessments; (3) assessing clinician and caregiver acceptance, usability, and perceived impact on care continuity; (4) examining the platform’s impact on time-to-detection of growth faltering and referral rates; and (5) analyzing cost-effectiveness and barriers to scale-up in low-resource settings. A mixed-methods design is employed. The quantitative component uses a prospective cohort of 1,200 children aged 0–5 years recruited from four primary care networks over 18 months, with follow-up at 3-month intervals. Data sources include digitally transmitted anthropometric measurements (length/height, weight, head circumference), standardized growth charts, nutritional intake logs, activity measures from wearables, and clinical outcomes. An artificial intelligence component develops a time-series forecasting model (LSTM or Prophet-based) augmented by gradient boosting to handle irregular sampling and missing data, with external validation on a separate 200-child cohort. Descriptive statistics summarize data completeness and measurement concordance between remote and in-clinic assessments. Inferential analyses employ Bland–Altman plots for measurement agreement, intraclass correlation coefficients for reliability, and regression analyses to identify predictors of growth trajectory deviations. The thematic qualitative component gathers data from semi-structured interviews with 25 clinicians and 40 caregivers, analyzed via thematic analysis to extract insights on usability, trust, and perceived impact on care coordination. The study adheres to ethical standards, with informed consent, data anonymization, and institutional review board approval. The expected findings include high concordance between remote and clinic measurements for weight and length/height in the majority of cases (limits of agreement within clinically acceptable ranges), substantial diagnostic accuracy of the AI growth-trajectory model in flagging potential faltering growth, and statistically significant improvements in time-to-intervention for at-risk children compared with historical controls. It is anticipated that the platform will demonstrate favorable usability scores from clinicians (System Usability Scale > 70) and positive caregiver acceptance (Net Promoter Score > 40), with qualitative data revealing enhanced perceived continuity of care and reduced travel burden. Cost-effectiveness analysis is expected to show favorable incremental cost per quality-adjusted life year gained when scaling the model to diverse settings, with key cost drivers identified as data transmission, device procurement, and training. The study contributes to knowledge by providing empirical evidence on the feasibility, accuracy, and clinical value of AI-enabled telemedicine for pediatric growth monitoring, offering a validated forecasting framework that integrates multi-source data, and detailing implementation considerations for low-resource environments. It also extends theoretical understanding of trust in AI-enabled pediatric care and the role of telemedicine in preventive child health, bridging health informatics, pediatric endocrinology, and health economics. The main conclusion anticipated is that AI-assisted pediatric telemedicine can reliably augment growth monitoring, shorten detection times for growth disturbances, and be cost-effective when accompanied by structured clinician training and caregiver engagement. Recommendations include scalable integration into national child health programs, development of standardized remote measurement protocols, robust data governance frameworks, and ongoing monitoring of equity impacts to ensure benefits across sociodemographic groups.