Richard Wen

Interventions implemented in geographic space (geo-interventions) have had success in reducing preventable deaths across the world. However, many studies supporting geo-interventions have focused on where to implement them rather than what they are. In this paper, we answer how to model and generate geo-interventions using spatial data, providing what these geo-interventions are and where to apply them. We defined geo-intervention modelling as a problem of optimizing actions and their locations, given the objective of maximizing predicted outcomes. To solve this, we produced a framework for transforming spatial data to model potential actions for generating geo-interventions. Finally, we conducted a case study of reducing traffic collisions in Toronto, Canada, to demonstrate the framework, which produced a machine learning model that discovered geo-interventions modifying red light camera, transit shelter, and wayfinding infrastructure predicted to reduce collisions by 5.7%. We highlight the importance of the framework for bridging research and practice through unified understanding, actionable outputs, human guidance, and iterative refinement. With recent advances in big data and artificial intelligence, we envision an acceleration in the discovery of geo-interventions and emergence of interdisciplinary work towards predicting accurate and precise future real-world outcomes at scale.

Interventions implemented in geographic space (geo-interventions), have had success in reducing preventable deaths across the world. However, many studies supporting geo-interventions have focused on where to implement them rather than what they are. In this paper, we answer how to model and generate geo-interventions using spatial data, providing what these geo-interventions are and where to apply them. We defined geo-intervention modelling as a problem of optimizing actions and their locations, given the objective of maximizing predicted outcomes. To solve this, we produced a framework for transforming spatial data to model potential actions for generating geo-interventions. Finally, we conducted a case study of reducing traffic collisions in Toronto, Canada, to demonstrate the framework, which produced a machine learning model that discovered geo-interventions modifying red light camera, transit shelter, and wayfinding infrastructure predicted to reduce collisions by 5.7%. We highlight the importance of the framework for bridging research and practice through unified understanding, actionable outputs, human guidance, and iterative refinement. With recent advances in big data and artificial intelligence, we envision an acceleration in the discovery of geo-interventions, and emergence of interdisciplinary work towards predicting accurate and precise future real-world outcomes at scale.

Background: Verbal autopsies (VAs) collect information on deaths in low and middle-income countries occurring outside healthcare facilities to estimate causes of death (CODs) for use in epidemiological or planning studies. Physician coding of VAs focused on the narrative of deaths and past symptoms is current best practice. Large language models (LLM) such as GPT-5 enable possible use of the narrative portion of VAs to assign CODs. However, there are few if any robust comparisons of LLMs to physician coding. Methods: We analyzed 6,939 VA records from a random sample of deaths in Sierra Leone (2019–2022) to compare five models: three LLMs (GPT-3.5, GPT-4, GPT-5) and two based on symptom algorithms (InterVA-5, InSilicoVA), against physician-assigned CODs. GPT models used narratives, whereas InterVA-5 and InSilicoVA relied on questionnaires. CODs were grouped into 19, 10, and 7 categories for adult, child, and neonatal deaths. We used cause specific mortality fraction (CSMF) accuracy and partial chance corrected concordance (PCCC) to assess population and individual-level agreement respectively, compared to the standard of physician coding. We stratified analyses by age group as CODs vary among neonates, children and adults. Results: Overall, GPT-5 outperformed all models (PCCC = 0.71), followed by GPT-4 (0.61), GPT-3.5 (0.56), InSilicoVA (0.44), and InterVA-5 (0.44). GPT-5 achieved the highest performance for adult (0.68), child (0.71), and neonatal (0.65) deaths. Across ages, performance increased from 1 month to 14 years and declined from 15 to 69 years. GPT-5, GPT-4, GPT-3.5, and InSilicoVA achieved the highest PCCC in 14, 7, 7, and 2 of the 30 CODs, respectively. At the population level, GPT-5 achieved the highest CSMF accuracy (0.9), while all other models had comparable performance (0.74–0.79). Conclusions: GPT models and InSilicoVA showed greater performance for specific CODs at the individual-level. GPT models demonstrated improvements over InterVA-5 and InSilicoVA models. This study provides foundational evidence for integrating LLM and algorithmic models with physician coding to improve the quality of VA data.

Many spatial decision support systems suffer from user adoption issues in practice due to lack of trust, technical expertise, and resources. Automated machine learning has recently allowed non-experts to explore and apply machine-learning models in the industry without requiring abundant expert knowledge and resources. This paper reviews recent literature from 136 papers, and proposes a general framework for integrating spatial decision support systems with automated machine learning as an opportunity to lower major user adoption barriers. Challenges of data quality, model interpretability, and practical usefulness are discussed as general considerations for system implementation. Research opportunities related to spatially explicit models in AutoML, and resource-aware, collaborative/connected, and human-centered systems are also discussed to address these challenges. This paper argues that integrating automated machine learning into spatial decision support systems can not only potentially encourage user adoption, but also mutually benefit research in both fields—bridging human-related and technical advancements for fostering future developments in spatial decision support systems and automated machine learning.

Many spatial decision support systems suffer from user adoption issues in practice due to lack of trust, technical expertise, and resources. Automated machine learning has recently allowed non-experts to explore and apply machine learning models in the industry without requiring abundant expert knowledge and resources. This paper reviews recent literature from 136 papers, and proposes a general framework for integrating spatial decision support systems with automated machine learning to lower major user adoption barriers. Challenges of data quality, model interpretability, and practical usefulness were discussed as general considerations for system implementation. Research opportunities related to spatially explicit models in AutoML, and resource-aware, collaborative/connected, and human-centered systems were also discussed to address these challenges. This paper argues that integrating spatial decision support systems with automated machine learning can not only encourage user adoption, but also mutually benefit research in both fields — bridging human-related and technical advancements for fostering future developments in spatial decision support systems and automated machine learning.

Since the rapid growth of urban populations, the study of urban systems has gained considerable attention from researchers, decision makers, governments, and organizations. Urban systems are complex and dynamic such that they produce emergent patterns such as self-organization and nonlinearity. Agent-based modelling presents an approach to simulating and abstracting urban systems to reveal and study emergent patterns from urban-related entities. However, agent-based models are difficult to effectively optimize and validate without high quality real-world data. Geosocial media data provides agent-based models with location-enabled data at high volumes and frequencies. Integrating agent-based models with geosocial media data presents opportunities in advancing and developing studies in urban systems. This paper provides a general overview of concepts, review of recent applications, and discussion of challenges and opportunities in the context of using geosocial media data in agent-based models for urban systems. We argue that ABMs focused on studying urban systems can benefit greatly from geosocial media data, given that research moves towards standard guidelines that enable the comparison and effective use of ABMs, and geosocial media data under appropriate circumstances and applications.

Verbal autopsies with physician assignment of cause of death (COD) are commonly used in settings where medical certification of deaths is uncommon. It remains unanswered if automated algorithms can replace physician assignment. We randomized verbal autopsy interviews for deaths in 117 villages in rural India to either physician or automated COD assignment. Twenty-four trained lay (non-medical) surveyors applied the allocated method using a laptop-based electronic system. Two of 25 physicians were allocated randomly to independently code the deaths in the physician assignment arm. Six algorithms (Naïve Bayes Classifier (NBC), King-Lu, InSilicoVA, InSilicoVA-NT, InterVA-4, and SmartVA) coded each death in the automated arm. The primary outcome was concordance with the COD distribution in the standard physician-assigned arm. Four thousand six hundred fifty-one (4651) deaths were allocated to physician (standard), and 4723 to automated arms. The two arms were nearly identical in demographics and key symptom patterns. The average concordances of automated algorithms with the standard were 62%, 56%, and 59% for adult, child, and neonatal deaths, respectively. Automated algorithms showed inconsistent results, even for causes that are relatively easy to identify such as road traffic injuries. Automated algorithms underestimated the number of cancer and suicide deaths in adults and overestimated other injuries in adults and children. Across all ages, average weighted concordance with the standard was 62% (range 79–45%) with the best to worst ranking automated algorithms being InterVA-4, InSilicoVA-NT, InSilicoVA, SmartVA, NBC, and King-Lu. Individual-level sensitivity for causes of adult deaths in the automated arm was low between the algorithms but high between two independent physicians in the physician arm. While desirable, automated algorithms require further development and rigorous evaluation. Lay reporting of deaths paired with physician COD assignment of verbal autopsies, despite some limitations, remains a practicable method to document the patterns of mortality reliably for unattended deaths. ClinicalTrials.gov , NCT02810366. Submitted on 11 April 2016.

Social media platforms allow millions of people worldwide to instantly share their thoughts online. Many people use social media to share traffic related experiences and events with online posts. A large amount of traffic related data can be obtained from these online posts – especially geosocial media data, where posts are tagged with geolocation information such as coordinates or place names. By extracting traffic events from geosocial media data, drivers can adapt to changing traffic conditions, while traffic management departments can propose timely and effective plans to improve traffic conditions. Most of the existing studies query traffic-related information based on a list of single keywords, which result in large amounts of noisy data – negative data containing one or more traffic-related keywords, but do not actually represent real-world traffic events. This paper aims to filter noisy data by mining association rules among words in positive data containing messages representing traffic events. Messages are more likely to be true traffic events if they follow the co-occurrence pattern of words mined from positive samples. A case study was conducted in Toronto, Canada using Twitter data. The tweets queried by the association rules were classified into non-traffic event, traffic accidents, roadwork, severe weather conditions, and special events with an 85% accuracy based on supervised machine learning methods. Compared with hourly average travel speed data, 81% of detected events were identified as real-world traffic events. This research sheds light on traffic condition monitoring in smart transportation platforms, which plays an important role for smart cities.