Background Health inequities exist between and within countries and communities for maternal and child health, nutrition, and early childhood development. Socially excluded bear the major brunt of this disparity. Innovative ways of providing healthcare are required to meet the needs of such populations. We report the development and feasibility testing of Sehat Ghar (Health House), an android-based digital application for volunteer health workers from a population not covered by Primary Healthcare (PHC). Methods We carried out a mixed-methods study with three steps. First, we conducted 13 in-depth interviews and two Focus Group Discussions (FGDs) with stakeholders to explore the gaps in community knowledge and practices. To address these gaps, we developed the Sehat Ghar App, comprised of video-based health education to demonstrate practices that mothers and families need to adopt. Second, we trained ten volunteer Community Health Workers (CHWs) from the same community to deliver health education using the App, and assessed their knowledge and skill improvement. Third, these CHWs visited pregnant and lactating mothers at home, that we monitored using a structured observation list. Results Initial exploration revealed a need for health-related knowledge and suboptimal healthcare utilization from neighbouring public hospitals. Sehat Ghar employed behaviour change techniques, including knowledge transfer, improving mothers' self-efficacy, and enhanced family involvement in mother and childcare to address this. Volunteer women were trained from the community, who, after the training, showed a significant improvement in mean knowledge score [Before: M = 8.00 (SD = 1.49), After: M = 11.40 (SD = 1.43), p=.0007]. Our monitoring found these CHWs excellent in their interaction with mothers and excellent or very good in using the App. The CHW and her community reported their liking and satisfaction with the App and wanted its delivery on a regular basis. Conclusions The digital application Sehat Ghar is a simple, easy-to-use resource for CHWs and is acceptable to the community. Mothers appreciate the content and presentation and are ready to incorporate its messages into their daily practices. The real-world effectiveness of the innovation is currently being tested on 250 mother-infant pairs. With its usefulness and adaptability, and the rapidly spreading mobile phone and Internet technology, the innovation can educate communities at a large scale in a minimum amount of time, contributing to equitable coverage of health services in resource-constrained settings.

Aim and objectives: The aim was to contribute to the editorial principles on the possible use of Artificial Intelligence (AI)- based tools for scientific writing. The objectives included: A. Enlist the inclusion and exclusion criteria to test ChatGPT use in scientific writing B. Develop evaluation criteria to assess the quality of articles written by human authors and ChatGPT C. Compare prospectively written manuscripts by human authors and ChatGPT Design: Prospective exploratory study Intervention: Human authors and ChatGPT were asked to write short journal articles on three topics: 1) Promotion of early childhood development in Pakistan 2) Interventions to improve gender-responsive health services in low-and-middle-income countries, and 3) The pitfalls in risk communication for COVID-19. We content analyzed the articles using an evaluation matrix. Outcome measures: The completeness, credibility, and scientific content of an article. Completeness meant that structure (IMRaD) and organization was maintained. Credibility required that others work is duly cited, with an accurate bibliography. Scientific content required specificity, data accuracy, cohesion, inclusivity, confidentiality, limitations, readability, and time efficiency. Results: The articles by human authors scored better than ChatGPT in completeness and credibility. Similarly, human-written articles scored better for most of the items in scientific content except for time efficiency where ChatGPT scored better. The methods section was absent in ChatGPT articles, and a majority of references in its bibliography were unverifiable. Conclusions: ChatGPT generates content that is believable but may not be true. The creators of this powerful model must step up and provide solutions to manage its glitches and potential misuse. In parallel, the academic departments, editors, and publishers must expect a growing utilization of ChatGPT and similar tools. Disallowing ChatGPT as a co-author may not be enough on their part. They must adapt the editorial policies, use measures to detect AI-based writing, and stop its likely implications for human health and life.