The algorithm journey map: a tangible approach to implementing AI solutions in healthcare.

When integrating AI tools in healthcare settings, complex interactions between technologies and primary users are not always fully understood or visible. This deficient and ambiguous understanding hampers attempts by healthcare organizations to adopt AI/ML, and it also creates new challenges for researchers to identify opportunities for simplifying adoption and developing best practices for the use of AI-based solutions. Our study fills this gap by documenting the process of designing, building, and maintaining an AI solution called SepsisWatch at Duke University Health System. We conducted 20 interviews with the team of engineers and scientists that led the multi-year effort to build the tool, integrate it into practice, and maintain the solution. This "Algorithm Journey Map" enumerates all social and technical activities throughout the AI solution's procurement, development, integration, and full lifecycle management. In addition to mapping the "who?" and "what?" of the adoption of the AI tool, we also show several 'lessons learned' throughout the algorithm journey maps including modeling assumptions, stakeholder inclusion, and organizational structure. In doing so, we identify generalizable insights about how to recognize and navigate barriers to AI/ML adoption in healthcare settings. We expect that this effort will further the development of best practices for operationalizing and sustaining ethical principles-in algorithmic systems.

Association of maternal nutritional status and small for gestational age neonates in peri-urban communities of Karachi, Pakistan: findings from the PRISMA study.

BACKGROUND: Early pregnancy nutritional status can be associated with adverse birth outcomes such as small-for-gestational age (SGA) and low birth weight (LBW). BMI (Body Mass Index) and MUAC (Mid-upper arm circumference) are easy to use assessments and are indicative of the pre-pregnancy nutritional status if obtained in the first trimester. This study primarily assesses the association of maternal nutritional status using BMI and MUAC with SGA in a community-based cohort of Pakistani women. It also aims to determine the predictive ability of MUAC and BMI in predicting SGA. Secondarily, we assessed the association between maternal nutrition and large for gestational age (LGA) and LBW. METHODS: This study is a secondary analysis of an ongoing pregnancy cohort "Pregnancy Risk Infant Surveillance and Measurement Alliance (PRISMA)"in Ibrahim Hyderi and Rehri Goth, Karachi. PRISMA participants who were enrolled between January 2021 to August 2022 were included given they had a gestational age < 14 weeks confirmed via ultrasound, MUAC and BMI measurements were available and birth weight was captured within 72 hours. Multivariable logistic regression was used to determine an association between maternal nutritional status and SGA. The PRISMA study was approved by the Aga Khan University Ethics Review Committee (2021-5920-15,518). RESULTS: Of 926 women included in the analysis, 26.6% (n = 247) had a low MUAC (< 23 cm) while 18.4% (n = 171) were underweight (BMI < 18.5 kg/m2). Nearly one third of low MUAC and underweight women delivered SGA infants (34.4 and 35.1% respectively). Underweight women and women with low MUAC had a statistically significant association with SGA (Underweight: OR 1.49, 95% CI 1.1,2.4; Low MUAC-OR 1.64, 95% CI 1.2,2.3) as well as LBW (Underweight: OR-1.63, 95% CI 1.1,2.4; Low MUAC-OR-1.63, 95% CI 1.2,2.3). ROC curves showed that MUAC and BMI had modest predictability for SGA (AUC < 0.7). CONCLUSION: Maternal nutritional status as indicated by BMI and MUAC are strongly associated with adverse pregnancy outcomes including SGA, LGA and LBW. Although MUAC and BMI are widely used to determine maternal nutritional status, they have poor predictive ability for newborn size. Further research is needed to identify other tools or a combination of tools to better predict adverse birth outcomes in resource-limited settings and plan interventions.

Applying the principles for digital development to improve maternal and child health in the Peri-urban areas of Karachi, Pakistan.

Low- middle-income countries, including Pakistan, are facing significant obstacles in their efforts to achieve the global targets for maternal, newborn, and child health (MNCH) defined by the Sustainable Development Goals. Barriers at the individual, health system, and contextual levels undermine healthcare access for pregnant women and children, disproportionately affecting those in low-resource settings. To address these challenges in the high-mortality, peri-urban areas of Karachi, VITAL Pakistan Trust and Aga Khan University launched a digital health intervention (DHI) to stimulate demand for health services and streamline care management for health workers at the primary care level. In this case study, we present a narrative review of the design, development, and deployment of the DHI, an Android-based application, in accordance with the Principles for Digital Development. We draw on the initial experience with implementation to reflect on how each of the nine Principles was considered during different phases of the project lifecycle, focusing on the lessons learned and challenges encountered during this process. By engaging with end-users and understanding the community, we were able to map existing relationships and workflows onto a digital platform to address major challenges hindering service delivery. Leveraging insights from field observations and user feedback, we collaborated with experts in healthcare and technology to develop the DHI, which has now scaled to 44 peri-urban settlements in Karachi. Our experience underscores the value of substantiated frameworks like the Principles. However, on-ground challenges reveal important caveats requiring further assessment. These include building community trust in new digital systems and ensuring the ethical use of health data, particularly in low digital and data literacy contexts. Based on this understanding, we share recommendations for conditions central to the effective integration and uptake of technology in healthcare, specifically within the context of digital health for MNCH.

Cohort profile: the Pregnancy Risk Infant Surveillance and Measurement Alliance (PRISMA) - Pakistan.

PURPOSE: Pakistan has disproportionately high maternal and neonatal morbidity and mortality. There is a lack of detailed, population-representative data to provide evidence for risk factors, morbidities and mortality among pregnant women and their newborns. The Pregnancy Risk, Infant Surveillance and Measurement Alliance (PRISMA) is a multicountry open cohort that aims to collect high-dimensional, standardised data across five South Asian and African countries for estimating risk and developing innovative strategies to optimise pregnancy outcomes for mothers and their newborns. This study presents the baseline maternal and neonatal characteristics of the Pakistan site occurring prior to the launch of a multisite, harmonised protocol. PARTICIPANTS: PRISMA Pakistan study is being conducted at two periurban field sites in Karachi, Pakistan. These sites have primary healthcare clinics where pregnant women and their newborns are followed during the antenatal, intrapartum and postnatal periods up to 1 year after delivery. All encounters are captured electronically through a custom-built Android application. A total of 3731 pregnant women with a mean age of 26.6±5.8 years at the time of pregnancy with neonatal outcomes between January 2021 and August 2022 serve as a baseline for the PRISMA Pakistan study. FINDINGS TO DATE: In this cohort, live births accounted for the majority of pregnancy outcomes (92%, n=3478), followed by miscarriages/abortions (5.5%, n=205) and stillbirths (2.6%, n=98). Twenty-two per cent of women (n=786) delivered at home. One out of every four neonates was low birth weight (<2500 g), and one out of every five was preterm (gestational age <37 weeks). The maternal mortality rate was 172/100 000 pregnancies, the neonatal mortality rate was 52/1000 live births and the stillbirth rate was 27/1000 births. The three most common causes of neonatal deaths obtained through verbal autopsy were perinatal asphyxia (39.6%), preterm births (19.8%) and infections (12.6%). FUTURE PLANS: The PRISMA cohort will provide data-driven insights to prioritise and design interventions to improve maternal and neonatal outcomes in low-resource regions. TRIAL REGISTRATION NUMBER: NCT05904145.

Can machine learning methods be used for identification of at-risk neonates in low-resource settings? A prospective cohort study.

INTRODUCTION: Timely identification of at-risk neonates (ARNs) in the community is essential to reduce mortality in low-resource settings. Tools such as American Academy of Pediatrics pulse oximetry (POx) and WHO Young Infants Clinical Signs (WHOS) have high specificity but low sensitivity to identify ARNs. Our aim was assessing the value of POx and WHOS independently, in combination and with machine learning (ML) from clinical features, to detect ARNs in a low/middle-income country. METHODS: This prospective cohort study was conducted in a periurban community in Pakistan. Eligible live births were screened using WHOS and POx along with clinical information regarding pregnancy and delivery. The enrolled neonates were followed for 4 weeks of life to assess the vital status. The predictive value to identify ARNs, of POx, WHOS and an ML model using maternal and neonatal clinical features, was assessed. RESULTS: Of 1336 neonates, 68 (5%) had adverse outcomes, that is, sepsis (n=40, 59%), critical congenital heart disease (n=2, 3%), severe persistent pulmonary hypertension (n=1), hospitalisation (n=8, 12%) and death (n=17, 25%) assessed at 4 weeks of life. Specificity of POx and WHOS to independently identify ARNs was 99%, with sensitivity of 19% and 63%,respectively. Combining both improved sensitivity to 70%, keeping specificity at 98%. An ML model using clinical variables had 44% specificity and 76% sensitivity. A staged assessment, where WHOS, POx and ML are sequentially used for triage, increased sensitivity to 85%, keeping specificity 75%. Using ML (when WHOS and POx negative) for community follow-up detected the majority of ARNs. CONCLUSION: Classic screening, combined with ML, can help maximise identifying ARNs and could be embedded in low-resource clinical settings, thereby improving outcome. Sequential use of classic assessment and clinical ML identifies the most ARNs in the community, still optimising follow-up clinical care.

Why do women still give birth at home; perceptions of Pakistani women and decision-makers from marginalized communities.

In low- and middle-income countries (LMICs), maternal and newborn mortality is high due to the high prevalence of home births. Understanding the reasons behind this behavior is essential for improving maternal and newborn outcomes. Therefore, a qualitative exploratory study was conducted in a peri-urban community in Karachi, Pakistan to understand the perceptions of pregnant women who delivered at home despite receiving antenatal care and the perceptions of their decision-makers regarding this behavior. In-depth interviews were conducted with 15 randomly sampled women who chose to deliver at home after receiving antenatal care at a health facility, as well as 15 family members who were purposively identified as decision-makers by the women themselves. Thematic analysis was performed to explore the perceptions, myths, and cultural beliefs about homebirths as well as women's decision-making power related to childbirth. The three main themes identified showed that traditional beliefs and practices, poverty and gender inequality, and poor healthcare systems significantly influence the preference for childbirth. Traditional beliefs and practices, including religious and cultural beliefs, played a role in perceiving childbirth as a natural process best managed at home. The presence of traditional birth attendants who provide personalized care and emotional support further reinforced this preference. Gender inequalities, including limited access to mobile phones and women's caregiving roles, were identified as barriers to seeking formal healthcare at the time of delivery. Additionally, poor experiences with the formal healthcare system, such as the poor attitude of formal healthcare workers and fear of medical interventions, also contributed to the decision to deliver at home. The study highlighted the complex interplay between traditional/religious beliefs, gender inequalities, and healthcare experiences in shaping the decision to deliver at home despite receiving ANC services in marginalized settings. Addressing these factors is necessary for promoting facility-based delivery and improving maternal and neonatal outcomes in LMICs.

Comparison of Masimo Total Hemoglobin SpHb® continuous non-invasive hemoglobin monitoring device with laboratory complete blood count measurement using venous sample: Protocol for an observational substudy of the Pregnancy Risk and Infant Surveillance and Measurement Alliance Maternal and Newborn Health (PRISMA MNH) study.

Background: The Masimo Total Hemoglobin SpHb® is a continuous and non-invasive handheld device to measure hemoglobin levels. Previous research has found that SpHb is able to accurately detect hemoglobin levels in adult patients with a similar degree of bias and standard deviation to point-of-care invasive method measurements. Generally, limited clinical evidence, lack of validation of Masimo at higher than and lower than hemoglobin threshold values, and scientific consensus supporting the use of Masimo for accurate hemoglobin testing for the diagnosis of anemia during pregnancy calls for further research. Methods and analysis: The proposed prospective cohort will be nested within the ongoing Pregnancy Risk and Infant Surveillance and Measurement Alliance (PRISMA) Maternal and Newborn Health (MNH) study. Three study sites (located in Zambia, Kenya, and Pakistan) will participate and collect hemoglobin data at five time points (<20 weeks, 20 weeks, 28 weeks, 36 weeks' gestation, and six weeks postpartum). We will measure hemoglobin using a venous blood sample via hematology auto-analyzer complete blood count (gold standard) and the non-invasive device. The primary objective is to assess agreement between Masimo total hemoglobin and complete blood count and on a continuous scale using Intraclass Correlation Coefficient and Bland-Altman Analysis. The second objective is to assess agreement between the two measures on a binary scale using Positive Percentage Agreement and Negative Percentage Agreement, Cohen's Kappa, and McNemar Test. On an ordinal scale, agreement will be measured using Weighted Cohen's Kappa and Harrel's Concordance Index. Lastly, we will assess factors that might affect the accuracy of Masimo total hemoglobin using linear mixed models. Conclusions: The primary aim of this study is to assess the validity of the non-invasive Masimo device compared to the gold standard method of invasive hemoglobin measurements during pregnancy and postpartum periods for the diagnosis of anemia.

Health data ecosystem in Pakistan: a multisectoral qualitative assessment of needs and opportunities.

OBJECTIVE: Data are essential for tracking and monitoring of progress on health-related sustainable development goals (SDGs). But the capacity to analyse subnational and granular data is limited in low and middle-income countries. Although Pakistan lags behind on achieving several health-related SDGs, its health information capacity is nascent. Through an exploratory qualitative approach, we aimed to understand the current landscape and perceptions on data in decision-making among stakeholders of the health data ecosystem in Pakistan. DESIGN: We used an exploratory qualitative study design. SETTING: This study was conducted at the Aga Khan University, Karachi, Pakistan. PARTICIPANTS: We conducted semistructured, in-depth interviews with multidisciplinary and multisectoral stakeholders from academia, hospital management, government, Non-governmental organisations and other relevant private entities till thematic saturation was achieved. Interviews were recorded and transcribed, followed by thematic analysis using NVivo. RESULTS: Thematic analysis of 15 in-depth interviews revealed three major themes: (1) institutions are collecting data but face barriers to its effective utilisation for decision-making. These include lack of collection of needs-responsive data, lack of a gender/equity in data collection efforts, inadequate digitisation, data reliability and limited analytical ability; (2) there is openness and enthusiasm for sharing data for advancing health; however, multiple barriers hinder this including appropriate regulatory frameworks, platforms for sharing data, interoperability and defined win-win scenarios; (3) there is limited capacity in the area of both human capital and infrastructure, for being able to use data to advance health, but there is appetite to improve and invest in capacity in this area. CONCLUSIONS: Our study identified key areas of focus that can contribute to orient a national health data roadmap and ecosystem in Pakistan.

Maternal and environmental Impact assessment on Neurodevelopment in Early childhood years (MINE): a prospective cohort study protocol from a low, middle-income country.

INTRODUCTION: Environmental and psychosocial adversities negatively impact children's developmental outcomes. When these factors are experienced in early childhood-a sensitive period of development-the developing brain can be altered. While these associations have been drawn in high-income countries, it is necessary to understand child growth, neurodevelopment, and the role of environmental factors in developmental trajectories in low-income settings. The objective of this study is to longitudinally assess how demographic factors, maternal health, maternal development, and child health, are related to child development on a behavioural, cognitive, and neuroimaging level in low-socioeconomic communities. METHODS AND ANALYSIS: Mother-child dyads will be identified in the peri-urban field sites of Rehri Goth and Ibrahim Hyderi, Karachi, Pakistan. Dyads will undergo yearly assessments for 4 years beginning when the child is 1 month, 3 months or 6 months of age (+≤30 days of age) (depending on group assignment). Maternal assessments include anthropometry, behavioural, cognitive, and developmental assessments (Edinburgh Postnatal Depression Scale; Parenting Stress Index; Maternal Autonomy Index; Hurt, Insult, Threaten, Scream Tool; Reynolds Intellectual Assessment Scales (RIAS)), and biological samples collection (breast milk, blood, stool, hair). Children's assessments include anthropometry, developmental assessments (Global Scales for Early Development (GSED); RIAS), MRI brain assessments, and biological sample collection (blood, stool, hair). Using cross-sectional and longitudinal data with statistical analysis tools, associations will be quantified between brain structure (MRI) and connectivity (resting state connectivity and diffusion tensor imaging), general cognitive skills (RIAS, GSED) and environmental influences (nutrition via biological samples, maternal mental health via questionnaires) through repeated measures analysis of variance tests and χ2 tests. Quantile regression and cortical analyses will be conducted to understand how demographic factors are related to the associations found. ETHICS AND DISSEMINATION: The study has received ethical approval from the Aga Khan University Ethics Review Committee. The study's findings will be disseminated through scientific publications and project summaries for the participants.

High prevalence of food insecurity in peri-urban communities in Karachi, Pakistan: The flood aftermath.

BACKGROUND: Food insecurity (FI) is at a steep risk, especially in low middle-income countries. FI is further compounded in areas that experience environmental and economic instability, thus a reassessment in such critical times is required for burden estimation and to propose targeted interventions. AIMS: The objectives of this study were to assess the prevalence of, and sociodemographic factors associated with FI, as well as the coping strategies utilized in response to FI in peri-urban communities in Karachi, Pakistan. METHODS: We conducted a cross-sectional survey from November-December 2022 on 400 households in four peri-urban communities in Karachi, Pakistan. The Household Food Insecurity Access Scale (HFIAS) and reduced Coping Strategies Index (rCSI) questionnaire were used to assess FI. A Poisson regression was used to assess associations between sociodemographic factors and FI. RESULTS: The overall prevalence of FI was found to be 60.2% (n = 241) of which 33.8% (n = 135) were severely food insecure. Age, women's and breadwinners' education, women's occupation, and parity were significantly associated with FI. Participants reported relying on less expensive foods (44%) and borrowing food or help from others (35%) as the most common coping strategies overall in the FI households. CONCLUSION: With more than half the households facing FI and adopting severe measures to cope in these communities, it is pivotal to design and test interventions that can withstand economic and climate catastrophes and help ensure a safety need for food security for the most vulnerable.

Pregnant Women's Experiences with Midwifery-Led Antenatal Care Services in Peri-Urban Communities in Karachi, Pakistan.

Purpose: To understand pregnant women's experience with midwifery-led antenatal care services using the Respectful Maternity Care charter in primary health centers in Karachi, Pakistan. Methods: This cross-sectional study was at Rehri Goth and Ibrahim Hyderi, two peri-urban communities in Karachi, Pakistan, where women receive antenatal care services. All pregnant women in their third trimester who consented during the study period were included. The participants were asked about access to care, antenatal care experience, person-centered approach, and general satisfaction with the facility using a pre-designed questionnaire. These themes were mapped onto the universal Respectful Maternity Care charter. Descriptive statistics were used to summarize the findings in each of these themes. Multivariable logistic regression techniques to determine the relationship between the dependent and independent variables. Results: There were 904 women who agreed to participate in this study during January to December 2021. Majority of the women (94%, n=854) were satisfied with the operating hours and cleanliness. More than 90% of the women reported positive experiences regarding privacy, respectful treatment by midwives, and non-discriminatory care. However, 40% (n=362) of the women reported not receiving adequate information and informed consent before a medical procedure, while 65% (n=587) reported poor counseling for birth preparedness. Maternal age, women's occupation, women's education, and parity were found to be significantly associated with respect provided, satisfaction with counseling and the consent process. Conclusion: This study reported satisfaction of pregnant women with the facility's ambiance, respect, and care; however, poor communication skills regarding consent and antenatal counseling were reported. The findings suggest the need for more efficient strategies, such as regular respectful maternity care and technical training to strengthen midwife-patient interactions and enhance overall satisfaction, thus improving maternal and newborn outcomes.

Evaluation of data analytics workshop using RStudio amongst medical students in Pakistan.

A quasi-experimental study was conducted at the Aga Khan University, Karachi, Pakistan, to evaluate the outcomes of a series of workshops on 25 medical students' statistical knowledge and acceptance of RStudio. The knowledge in each of the five sessions was assessed using pre- and post- knowledge-based quizzes. In addition, the Student's Attitude Towards Statistics (SATS-36) and the Technology Acceptance Model were used. Data analysis on RStudio revealed a statistically significant improvement in knowledge in all five sessions (p<0.05). SATS-36 showed statistically significant improvement in Cognitive Competence (p<0.001). RStudio had commendable acceptance with relatively high scores of Attitudes (behavioural intention, median = 6.00 [5.20-7.00]) and Utility (perceived usefulness, median = 5.20 [4.10-6.20]). In conclusion, medical students had improved statistical knowledge and acceptance towards the novel statistical tool. Hence, further studies must evaluate the effectiveness of RStudio when integrated as part of the medical curriculum.

The Impact of Maternal Preeclampsia and Hyperglycemia on the Cardiovascular Health of the Offspring: A Systematic Review and Meta-analysis.

OBJECTIVES: The objective of this review was to assess the impact of maternal preeclampsia or hyperglycemia on the body composition and cardiovascular health in the offspring. STUDY DESIGN: We conducted a systematic review utilizing PubMed, EBSCO, CINAHLPlus, Cochrane Library, and Web of Science to include all studies assessing the impact of preeclampsia/eclampsia and/or gestational/pregestational diabetes mellitus on the health of the offspring (children <10 years of age). The health measures included anthropometry, cardiac dimensions and function, and vascular function. We performed a meta-analysis using Review Manager software and computed net risk ratio (RR) with 95% confidence interval (CI) for dichotomous data and mean difference (MD) with 95% CI for continuous data. RESULTS: There were 6,376 studies in total, of which 45 were included in the review and 40 in the meta-analysis. The results demonstrated higher birth weight (MD: 0.12 kg; 95% CI: 0.06-0.18) and systolic and diastolic blood pressure (BP; MD: 5.98 mm Hg; 95% CI: 5.64-6.32 and MD: 3.27 mm Hg; 95% CI: 0.65-5.89, respectively) in the offspring of mothers with gestational diabetes compared to controls. In contrast, the offspring of mothers with preeclampsia had lower birth weight (MD: -0.41 kg; 95% CI: -0.7 to -0.11); however, they had increased systolic (MD: 2.2 mm Hg; 95% CI: 1.28-3.12) and diastolic BP (MD: 1.41 mm Hg; 95% CI: 0.3-2.52) compared to controls. There is lack of data to conduct a meta-analysis of cardiac morphology, functional, and vascular imaging parameters. CONCLUSION: These findings suggest that the in-utero milieu can have a permanent impact on the body composition and vascular health of the offspring. Future work warrants multicenter prospective studies to understand the mechanism and the actual effect of exposure to maternal hyperglycemia and high BP on the cardiovascular health of the offspring and long-term outcomes. KEY POINTS: · Adverse in-utero exposures may have an impact on cardiovascular risk in children.. · Maternal hyperglycemia/preeclampsia lead to changes in birthweight and BP.. · Limited echocardiographic and vascular imaging data in these cohorts necessitates future work..

Neurodevelopment assessment of small for gestational age children in a community-based cohort from Pakistan.

BACKGROUND: Children born small for gestational age (SGA) may experience more long-term neurodevelopmental issues than those born appropriate for gestational age (AGA). This study aimed to assess differences in the neurodevelopment of children born SGA or AGA within a periurban community in Pakistan. METHODS: This was a prospective cohort study in which study participants were followed from the pilot Doppler cohort study conducted in 2018. This pilot study aimed to develop a pregnancy risk stratification model using machine learning on fetal Dopplers. This project identified 119 newborns who were born SGA (2.4±0.4 kg) based on International Fetal and Newborn Growth Consortium standards. We assessed 180 children (90 SGA and 90 AGA) between 2 and 4 years of age (76% of follow-up rate) using the Malawi Developmental Assessment Tool (MDAT). FINDINGS: Multivariable linear regression analysis comparing the absolute scores of MDAT showed significantly lower fine motor scores (ß: -0.98; 95% CI -1.90 to -0.06) among SGAs, whereas comparing the z-scores using multivariable logistic regression, SGA children had three times higher odds of overall z-scores ≤-2 (OR: 3.78; 95% CI 1.20 to 11.89) as compared with AGA children. INTERPRETATION: SGA exposure is associated with poor performance on overall MDAT, mainly due to changes in the fine motor domain in young children. The scores on the other domains (gross motor, language and social) were also lower among SGAs; however, none of these reached statistical significance. There is a need to design follow-up studies to assess the impact of SGA on child's neurodevelopmental trajectory and school performance.

Diagnostic accuracy of artificial intelligence for detecting gastrointestinal luminal pathologies: A systematic review and meta-analysis.

Background: Artificial Intelligence (AI) holds considerable promise for diagnostics in the field of gastroenterology. This systematic review and meta-analysis aims to assess the diagnostic accuracy of AI models compared with the gold standard of experts and histopathology for the diagnosis of various gastrointestinal (GI) luminal pathologies including polyps, neoplasms, and inflammatory bowel disease. Methods: We searched PubMed, CINAHL, Wiley Cochrane Library, and Web of Science electronic databases to identify studies assessing the diagnostic performance of AI models for GI luminal pathologies. We extracted binary diagnostic accuracy data and constructed contingency tables to derive the outcomes of interest: sensitivity and specificity. We performed a meta-analysis and hierarchical summary receiver operating characteristic curves (HSROC). The risk of bias was assessed using Quality Assessment for Diagnostic Accuracy Studies-2 (QUADAS-2) tool. Subgroup analyses were conducted based on the type of GI luminal disease, AI model, reference standard, and type of data used for analysis. This study is registered with PROSPERO (CRD42021288360). Findings: We included 73 studies, of which 31 were externally validated and provided sufficient information for inclusion in the meta-analysis. The overall sensitivity of AI for detecting GI luminal pathologies was 91.9% (95% CI: 89.0-94.1) and specificity was 91.7% (95% CI: 87.4-94.7). Deep learning models (sensitivity: 89.8%, specificity: 91.9%) and ensemble methods (sensitivity: 95.4%, specificity: 90.9%) were the most commonly used models in the included studies. Majority of studies (n = 56, 76.7%) had a high risk of selection bias while 74% (n = 54) studies were low risk on reference standard and 67% (n = 49) were low risk for flow and timing bias. Interpretation: The review suggests high sensitivity and specificity of AI models for the detection of GI luminal pathologies. There is a need for large, multi-center trials in both high income countries and low- and middle- income countries to assess the performance of these AI models in real clinical settings and its impact on diagnosis and prognosis. Systematic review registration: [https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=288360], identifier [CRD42021288360].

Health workers' experience of a digital health intervention implemented in peri-urban communities in Karachi, Pakistan.

Objective: Digital health interventions (DHIs) have the potential to improve access and quality of care in low-middle-income countries. The aim of this study was to assess the acceptability, usability and aesthetics of a DHI by frontline workers in peri-urban community settings in Karachi, Pakistan. Methods: A mixed-methods study was carried out in peri-urban field sites in Karachi, Pakistan, where maternal and childcare services are provided through front-line care providers using a DHI. These workers include community health workers, midwives, and physicians who were using the DHI for at least six months. For quantitative data, a questionnaire regarding the module design and interface, technical difficulty, and appropriate utilisation was assessed using a 5-point Likert scale. For qualitative data, focus group discussions (FGDs) based on experiences regarding operability, design, its effect on work efficiency and the provision of beneficial health services were conducted. Results: There were 93 respondents for the quantitative questionnaire who reported high satisfaction (>85%) with the DHI in many themes including content quality, aesthetics and ease of use. Participants were least satisfied with service quality (45% satisfaction only) due to issues related to data sync and network connections in these areas. During the FGDs, the workers stated that the DHI helped them with accessing previous data and providing quality health care services to the community. Conclusion: Although frontline workers reported a few technical difficulties while using the DHI, the majority reported that it was acceptable, had user-friendly features and was beneficial in their work processes.

Impact of COVID-19 pandemic on research and careers of early career researchers: a DOHaD perspective.

The COVID-19 pandemic has exposed several inequalities worldwide, including the populations' access to healthcare systems and economic differences that impact the access to vaccination, medical resources, and health care services. Scientific research activities were not an exception, such that scientific research was profoundly impacted globally. Research trainees and early career researchers (ECRs) are the life force of scientific discovery around the world, and their work and progress in research was dramatically affected by the COVID-19 pandemic. ECRs are a particularly vulnerable group as they are in a formative stage of their scientific careers, any disruptions during which is going to likely impact their lifelong career trajectory. To understand how COVID-19 impacted lives, career development plans, and research of Developmental Origins of Health and Disease (DOHaD) ECRs, the International DOHaD ECR committee formed a special interest group comprising of ECR representatives of International DOHaD affiliated Societies/Chapters from around the world (Australia and New Zealand, Canada, French Speaking DOHaD, Japan, Latin America, Pakistan and USA). The anecdotal evidence summarized in this brief report, provide an overview of the findings of this special interest group, specifically on the impact of the evolving COVID-19 pandemic on daily research activities and its effects on career development plans of ECRs. We also discuss how our learnings from these shared experiences can strengthen collaborative work for the current and future generation of scientists.

Maternal and child surveillance in peri-urban communities: Perceptions of women and community health workers from Pakistan.

Community health workers (CHWs) in maternal, newborn, and child health (MNCH) programs play an important role in demographic surveillance activities; however, there is lack of literature regarding the community and CHWs' perceptions about these activities. The purpose of this study was to explore perceptions of married women of reproductive age (MWRA) regarding the role of CHWs involved in maternal and child surveillance and explore facilitators and barriers for CHWs involved in surveillance activities. A qualitative study was conducted in five peri-urban surveillance sites along the coastal belt of Bin Qasim Town, Karachi, Pakistan. In-depth interviews were conducted with 25 randomly selected MWRAs and 15 CHWs. A thematic analysis was performed to explore perceptions, barriers, and facilitators of the study participants about maternal and child surveillance activities. The results showed that MWRAs perceived surveillance CHWs as service providers with regards to standard counselling i.e. importance of antenatal care, nutrition, immunization, and distribution of iron and folic acid tablets to pregnant women, child growth assessment, and referral of sick children to the health facility. Trust in the CHWs was an enabler for MWRAs, whereas lack of incentives was cited as a barrier to share their health data. CHWs perceived themselves as a bridge in liaising community with the primary health care facility. They highlighted an enabling environment such as appreciation, supportive supervision, training, and utilization of digital data collection tools as facilitators for their work. Low health literacy of the communities, lack of provision of incentives by CHWs to the community, and facility-based experiences of the community were reported as barriers. Surveillance CHWs are an integral link between the health facility and MWRAs. Hence an enabling environment may lead to improved health service delivery, translating into meaningful impact for the mother and child.

Deep learning applications in neuro-oncology.

Deep learning (DL) is a relatively newer subdomain of machine learning (ML) with incredible potential for certain applications in the medical field. Given recent advances in its use in neuro-oncology, its role in diagnosing, prognosticating, and managing the care of cancer patients has been the subject of many research studies. The gamut of studies has shown that the landscape of algorithmic methods is constantly improving with each iteration from its inception. With the increase in the availability of high-quality data, more training sets will allow for higher fidelity models. However, logistical and ethical concerns over a prospective trial comparing prognostic abilities of DL and physicians severely limit the ability of this technology to be widely adopted. One of the medical tenets is judgment, a facet of medical decision making in DL that is often missing because of its inherent nature as a "black box." A natural distrust for newer technology, combined with a lack of autonomy that is normally expected in our current medical practices, is just one of several important limitations in implementation. In our review, we will first define and outline the different types of artificial intelligence (AI) as well as the role of AI in the current advances of clinical medicine. We briefly highlight several of the salient studies using different methods of DL in the realm of neuroradiology and summarize the key findings and challenges faced when using this nascent technology, particularly ethical challenges that could be faced by users of DL.