World Bank Development Economics’ Post

#ResearchSpotlight 👩💻: In the quest to provide targeted aid for poverty alleviation, survey data is no longer the only option. 📈 🔎 Teaming up with IBM, our AI and Innovation Center of Excellence's geospatial analytics team utilized machine learning and fused diverse open source data for a more holistic, frequent, and accurate view. We have woven demographics, remote sensing, environmental, and points of interests data via OpenStreetMap to craft a richer, more frequent, and precise perspective on wealth levels in specific areas of the Philippines. 🤖 Together, we're pioneering the application of geospatial data and machine learning for investment planning and infrastructure rollout in the country. 🌇 🎉 We are also thrilled to announce that our research paper, showcasing a model that builds upon previous approaches for wealth level estimation, won the Best Paper award at the GeoSocial 2023 workshop of the prestigious conference ACM SIGSPATIAL 2023. 🇩🇪 SIGSPATIAL is an annual event that brings together leading minds in geospatial data science and geographic information systems. 🗺️ 🌐 If you're interested in learning more about our research and its potential implications, you can visit the GeoSocial 2023 workshop website (https://hubs.la/Q029R5wn0) or reach out to us directly: https://hubs.la/Q029R8rG0 #GeoSocial2023 #SIGSPATIAL2023 #data #innovation #machinelearning #ai #ai4good #GeospatialData #PovertyAlleviation

📚 Research Paper Publication: 📌 AIHC-2023 | Bratislava, Slovakia | October 24-26, 2023 📅 Springer Journal Publication. 👥 Authors: YAGNESH CHALLAGUNDLA, K BADRI NARAYANAN, @Krishna Sai Devatha, @Sachi Nandan Mohanty 🚀 Description: In this study, we developed a user-friendly calorie tracking program that achieved 97.88% accuracy using a multi-model machine learning approach. Leveraging a Kaggle dataset with information from over 16,000 individuals. Our methodology involved comprehensive steps, including data collection, preprocessing, dataset splitting, cross-validation, model selection, and evaluation, supported by data visualization techniques for interpretation. The findings not only indicate the effectiveness of the multi-model strategy in predicting calorie burn but also contribute to advancements in user-friendly applications for fitness and health objectives. Despite acknowledged limitations, the study sets a foundation for future improvements in calorie tracking and emphasizes the potential of machine learning approaches in health-related monitoring. Hashtag: #AIHealthcare #MachineLearning #FitnessMonitoring 🌐📊🩺

【Particle Swarm Optimization-Based Hyperparameters Tuning of Machine Learning Models for Big COVID-19 Data Analysis】 Full article: https://lnkd.in/gK2GUjiG (Authored by Hend S. Salem, et al., from Suez Canal University, Egypt.) #Machine_learning (ML) techniques have shown promise in analyzing complex data patterns and predicting disease outcomes. This work uses the Particle Swarm Optimization (#PSO) algorithm to effectively search the hyperparameter space and improve the predictive power of the machine learning models by identifying the optimal hyperparameters that can provide the highest accuracy, and develops and evaluates optimized ML models for predicting COVID-19 risk by using a dataset with a variety of clinical and epidemiological characteristics linked to COVID-19 cases. #Hyperparameter_Optimization #Computational_Intelligence

🔄 📊This study by Jannik Rößler and Detlef Schoder highlights the importance of choosing the proper method to predict the incremental effect of a treatment, especially in areas such as churn management and patient care. Comparing 15 "uplift modeling" and "heterogeneous treatment effects" (HTE) methods on synthetic and real-world datasets, the results reveal more robust and effective methods, including "Uplift-CHI" (U-CHI), " Uplift-Kullback-Leibler" (U-KL), "Bayesian Causal Forest" (BCF) and "Contextual Treatment Selection" (CTS). The conclusion emphasizes the importance of evaluating methods from both areas to optimize segmentation policies. 🔍 Methods compared include CHAID decision trees, Two-Model Approach, S-learner, Lai's Generalization, X-learner, Treatment Dummy Approach, R-learner, Uplift Random Forest, Contextual Treatment Selection, Interaction Tree, Causal Inference Tree, Generalized Random Forest and Bayesian Causal Forest. The analysis highlights the variable performance of these methods across different data sets, highlighting the need to consider both approaches to evaluate the incremental effect of treatment. The results indicate that improving segmentation policies requires the evaluation of several approaches from both research areas. #machinelearning #ML #research #data #datascientists #datascience #causality #causalinference #causalml #uplift #upliftmodeling #HTE #HeterogeneousTreatmentEffects #artificialintelligence #causalAI #econometrics

I was glad to give a talk in the workshop on 𝐒𝐩𝐚𝐜𝐞 𝐃𝐚𝐭𝐚 𝐀𝐧𝐚𝐥𝐲𝐭𝐢𝐜𝐬 at Indian Institute of Information Technology, SriCity 🇮🇳, sponsored by Indian Science and Engineering Research Board (SERB), kindly invited by Arun P.. I presented an overview of current challenges in #AI4EO and the various solutions we explore at European Space Agency - ESA #philab: 👉🏼 Physics-aware Machine Learning, 👉🏽 Multimodal AI, 👉🏾 Frugal Learning (semi-sup, active, etc.), 👉🏿 and now Geo-spatial Foundational Models! #Phileo

UNDP analysis of Human Development Index trends covering 135 countries using 40 years of data. Here is what they found: "While we confirm the existence of continued divergence in per capita income, we find the inverse for HDI. We find no statistically significant correlation between growth and non-income HDI improvements over a forty year period." That report was produced exactly 13 years ago. Have we learned anything? Have we changed anything? More chatbots? Technology enabling a new front end on the same backend models is not change. We need innovation that reassess traditional assumptions designed for an earlier era and move boldly forward with low-income communities, not private industry. We need a public health tech revolution. UNDP report: https://lnkd.in/gPwjFg-d

Artificial Intelligence is being introduced in public health in Eritrea. by engineer nguse berhane ! Artificial Intelligence (AI) has the potential to improve the capacity of public health to promote health in all communities. The paper presented at the seminar “Strengthening Public Health Research for Tomorrow's Challenges” shows the level of awareness and maturity of Eritrean students. Mulugeta Russom, PhD Mulugheta T. SOLOMON, PhD Filmon N. Yohannes M.D. Lidya Mussie Weldehawaryat Mesob St Eritrean Healthcare Professionals Network

Title: "Revolutionizing #Healthcare: Predicting #Strokes with A.I Machine Learning" Problem Statement: In a world grappling with health challenges, strokes remain a pervasive threat. Conventional methods often fall short in providing accurate predictions for early intervention. This year, my data science project aims to transform stroke prevention by employing cutting-edge machine learning techniques. Explore how we're addressing the critical need for precise and early stroke detection. Solution Description: Discover the future of healthcare with my machine learning-driven stroke prediction model. By analyzing a comprehensive dataset encompassing medical history, lifestyle factors, and genetic markers, our solution goes beyond traditional approaches. Uncover subtle patterns indicative of potential stroke risks, empowering individuals and healthcare professionals with early insights. Join us in leveraging data science for a healthier tomorrow - where proactive measures redefine stroke prevention. Test the Application using: https://lnkd.in/dFK5FxQp For those who want to watch the full video use this linke: https://lnkd.in/dKaCmfx3 Inspired by Liberty Health, Ministry of Health Zambia, IHM , World Health Organization, Right to Care and Google commitment to improving healthcare. Check my other projects: https://lnkd.in/dbZKcQPC #DataScience #MachineLearning #HealthTech #PredictiveAnalytics #StrokePrevention #HealthcareInnovation #AIinHealthcare #PublicHealth #TechForGood #DigitalHealth #DataDrivenDecisions #PreventiveHealth #HealthData #PrecisionMedicine #MedicalResearch #ai

Our paper titled 'Decision Tree Clustering for Time Series Data: An Approach for Enhanced Interpretability and Efficiency' has been accepted for PRICAI 2023. I will be attending the conference to present the paper next week.

📢 The UN-GGIM Thematic Networks today present “Geospatial data, analytics and GeoAI Accelerating the SDGs and Impacting National Priorities” The side-event aims to present the progress of the four UN-GGIM Thematic Networks during the inter-sessional period on different aspects of geospatial data and information, Geo-AI, and harnessing technologies for impacting countries and to support accelerating the 2030 Agenda and its 17 Sustainable Development Goals (SDGs). The objectives include to consider: - the role of geospatial data, technology and services to advance and accelerate the implementation of national priorities and global agendas. - current technological advancements and changes caused by artificial intelligence; and - the needs and progress towards syllabi and body of knowledge on GeoAI. Agenda 📜 Setting the scene Mr. Stefan Schweinfest, Director of the United Nations Statistics Division (UNSD) Department of Economic and Social Affairs (DESA) Short introduction, presentation and panel members Ms. Maria Brovelli Chair, UN-GGIM Academic Network Ms. Céline Rozenblat Chair, UN-GGIM Geospatial Societies Mr. Zaffar Sadiq Mohamed-Ghouse Chair, UN-GGIM Private Sector Network Mr. Alexandre Caldas Chair, UN Geospatial Network Moderated panel discussion Moderator: Ms. Ingrid Vanden Berghe National Geographic Institute Kingdom of Belgium Co-Chair – Committee of Experts on Global Geospatial Information Management