235 - Feasibility and Usability of Saving Little Lives Assist (SLL-Assist): A Novel AI Application to Improve Access to Content on Maternal and Newborn Health Care

Publication Number: 1223.235

Znabu Hadush. Kahsay, Mekelle University, Mekelle City, Tigray, Ethiopia; Araya Abrha. Medhanyie, Mekelle University, Mekelle, Tigray, Ethiopia; Simret Niguse. Weldebirhan, Mekelle University, Kigali, Kigali, Rwanda; Shishay W. Gebregiyorgis, Mekelle University, Mekelle, Tigray, Ethiopia; Haben H. Gebremichael, Mekelle University College of Health Sciences, Mekelle, Tigray, Ethiopia; Sushma Kallam, Rural Health Link Inc, Seattle, WA, United States; Teja Basireddy, Rural Health Link Inc., Visakhapatnam, Andhra Pradesh, India; Sara K. Berkelhamer, University of Washington, Seattle, WA, United States

Background: Maternal and neonatal mortality rates remain critically high in low resource settings (LRS). Outcomes are impacted by provider knowledge and access to up-to-date guidelines. Artificial intelligence (AI) is increasingly recognized as a transformative force for personalized access to educational and clinical content. A multi-disciplinary team created Saving Little Lives (SLL)-Assist -- an AI-driven application designed to support access to curated written and visual content on maternal and newborn care including materials from SLL, the WHO, Helping Mother and Babies Survive, and the Global Health Media Project. While audio read back, language adaptation and embedded video are included to optimize function, the feasibility and usability of SLL-Assist remain unknown.

Objective: To determine the feasibility and usability of an AI-based application designed to support access to content on maternal and newborn health amongst health care providers (HCP) in Tigray, Ethiopia.

Design/Methods: HCPs from 4 hospitals participated in a facilitated evaluation of SLL-Assist. Research staff helped install and orient providers and encouraged independent use with thematic suggestions weekly for 6w. Feedback was collected via WhatsApp, weekly meetings, and survey using a Post-Study System Usability Questionnaire (PSSUQ) and System Usability Scale (SUS). The PSSUQ evaluated three areas (usefulness, information quality and interface quality) with six statements to score on a 7-point Likert scale, resulting in a range from 6 (most satisfied) to 42 (least satisfied) for each.

Results: Out of 203 HCPs approached, 180 participated with an 88.6% response rate. The majority were female (73.4%), midwives (58%) or nurses (30%) with a mean age of 34.0 (+6.2) and 9.7 (+6.1) years of experience. HCPs reported high rates of functionality, confidence, uptake and ease of use (Fig1). Favorable scores on the PSSUQ included 12.5 for usefulness, 11.8 for information quality and 13.7 for interface quality (Fig 2). Participants found the information clear (98%), easy to locate (92%) and the majority agreed they would use the product frequently (88%). 70% felt they needed to learn more before using the App while 50% shared they needed technical support.

Conclusion(s): Use of a purpose-designed AI-driven application for accessing health care content was well received in a LRS. SLL-Assist represents a highly adaptable approach to improving HCP capacities through access to educational content. Adaptations to address offline function and local language will extend the impact of SLL-Assist. Future plans include larger scale implementation and evaluation.

Figure 1: PSSUQ Scores for SLL-Assist
Figure 1 PAS.pdfPSSUQ scores for usefulness, information quality and interface quality. Scores represent combined mean for 6 questions in each area, with scoring on a 7 point Likert scale where 1 is favorable and 7 is not. The range for 6 questions would be the lowest, most favorable score of 6 and least favorable of 42.

Figure 2: Post Use Scores for SSL-Assist
Figure 2 PAS.pdfThe majority of HCPs were in agreement around the ease of use and that they would like to use the application frequently.