Dengue fever presents a substantial challenge in developing countries where sanitation infrastructure is inadequate. The absence of comprehensive healthcare systems exacerbates the severity of dengue infections, potentially leading to life-threatening circumstances. Rapid response to dengue outbreaks is also challenging due to limited information exchange and integration. While timely dengue outbreak forecasts have the potential to prevent such outbreaks, the majority of dengue prediction studies have predominantly relied on data that impose significant burdens on individual countries for collection. In this study, our aim is to improve health equity in resource-constrained countries by exploring the effectiveness of high-resolution satellite imagery as a nontraditional and readily accessible data source. By leveraging the wealth of publicly available and easily obtainable satellite imagery, we present a scalable satellite extraction framework based on Sentinel Hub, a cloud-based computing platform. Furthermore, we introduce DengueNet, an innovative architecture that combines Vision Transformer, Radiomics, and long short-term memory to extract and integrate spatiotemporal features from satellite images. This enables dengue predictions on an epidemiological-week basis. To evaluate the effectiveness of our proposed method, we conducted experiments in five municipalities in Colombia. We utilized a dataset comprising 780 high-resolution Sentinel-2 satellite images for training and evaluation purposes. The performance of DengueNet was assessed using the mean absolute error (MAE) metric. Across the five municipalities, DengueNet achieved an average MAE of 43.92±42.19. Notably, the highest MAE was recorded in Cali at 113.65±0.08, whereas the lowest MAE was observed in Ibagué, amounting to 5.67±0.18. Our findings strongly support the efficacy of satellite imagery as a valuable resource for dengue prediction, particularly in informing public health policies within low- and middle-income countries. In these countries, where manually collected data of high quality is scarce and dengue virus prevalence is severe, satellite imagery can play a crucial role in improving dengue prevention and control strategies.

DengueNet: Dengue Prediction using Spatiotemporal Satellite Imagery for Resource-Limited Countries

Access to clean potable water is a fundamental re-
quirement for human well-being. However, numer-
ous challenges arise in water delivery and manage-
ment, particularly in the context of global climate
change. The increasing severity of droughts, social
inequalities, and infrastructure constraints necessi-
tates a concerted effort towards water management.
Despite the rise of Social Media as a platform for
societal engagement across various topics, the in-
clusion of water concerns in this discourse remains
remarkably underutilized, undeveloped, and under-
analyzed.
This paper aims to highlight the untapped oppor-
tunities offered by social media as a valuable tool
for water management. We emphasize the potential
of social media to serve as a versatile medium for
collecting and disseminating information regarding
water-related issues. By leveraging advanced AI
and Machine Learning techniques, we conduct tex-
tual analysis on a comprehensive dataset of water
utility tweets in the United States, extracting perti-
nent and meaningful insights. These insights serve
as a starting point for further advancements in wa-
ter management.

Harnessing Social Media Analysis to Enhance Water Management Efficiency

Geospatial technologies are increasingly important for various applications worldwide, including vegetation monitoring. Collecting ground truth data for specific geospatial tasks are challenging and time-consuming. Recently, the foundation model research have been explored, then pre-training on large-scale data and fine-tuning for specific tasks are important components of this technique. Although this approach can enhance the performance in downstream tasks such as satellite image translation, directly fine-tuning models pre-trained on natural images like ImageNet is suboptimal for geospatial data due to the inherent domain differences. In this paper, we propose a novel image translation approach with pre-training on geospatial-specific data and data augmentation. We present a case study where our method achieved outstanding results in a competition for inferring the Normalized Difference Vegetation Index from satellite-based Synthetic Aperture Radar data of cabbage farms. Our approach outperformed other methods with a 31% higher score than the second-ranked team and a 44% higher score than the average of the top five teams.

Pre-Training and Fine-Tuning for Satellite Image-to-Image Translation

We study restless multi-armed bandits (RMABs) in the context of public health, where there is a need to optimize resource allocation decisions. Until now, RMABs typically solve for the optimal planning policy by assuming  the reward function in the problem is fully known. However, in this work, we aim to study whether we can learn the most optimal rewards for an RMAB problem given some demonstrated, ideal behavior. To achieve this, we turn to inverse reinforcement learning (IRL) which is a field of study motivated by the desire to understand and learn the underlying reward structure of an agent's observed behavior. Existing IRL approaches predominantly focus on single agent systems, presenting limitations in dealing with the expansive state spaces characteristic of public health scenarios, where tens of thousands of arms are active simultaneously. We propose a new IRL algorithm specifically for RMAB settings that uses techniques from decision focused learning (DFL) to directly optimize the objective function, allowing for efficient and accurate updates to the learned rewards. We compare our algorithm with the max entropy IRL baseline on runtime and accuracy and find that our algorithm performs better on both metrics. We also propose a framework for how to apply this algorithm in the public health domain where expert trajectories come from domain experts.

Inverse Reinforcement Learning for Restless Multi-Armed Bandits with Application to Maternal and Child Health

 Current air quality surveillance ecosystems cannot monitor air pollution levels across time and space continuously, and lack the ability to integrate historical data from heterogeneous sources into artificial intelligence algorithms capable of providing more insights on the impact of severe air pollution. Integrating in-situ sensor and satellite data has attracted recent attention, but the characteristics of graph-based learning with heterogeneous air quality data have not been explored in depth. 
   
   In this paper, we present the PANORAMA knowledge graph based prediction approach of the links between pollutants exposure and health outcomes. PANORAMA aims at integrating heterogeneous air quality and health data in order to better predict and explain their relationships. An Extract, Transform, and Load process is used to design and incorporate discrete data into the knowledge graph, and a knowledge graph embedding model is trained and tested to assess inferential capabilities of the graph. A use case with a dataset related to the Gironde department in France is showcased. The promising preliminary results of a 31.376 MR, 0.312 MRR, and 0.254, 0.335, and 0.524 Hits@ 1, 3, and 10 respectively on a TransE model suggest the potential of leveraging semantic structuring and knowledge graph technology for environmental public health. 

IJCAI 2023 Workshop BridgeAICCHE Submissions

DengueNet: Dengue Prediction using Spatiotemporal Satellite Imagery for Resource-Limited Countries

Harnessing Social Media Analysis to Enhance Water Management Efficiency

Pre-Training and Fine-Tuning for Satellite Image-to-Image Translation

Inverse Reinforcement Learning for Restless Multi-Armed Bandits with Application to Maternal and Child Health

The PANORAMA knowledge graph based prediction of outdoor air quality impact on health