Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Can Explainable AI Assess Personalized Health Risks from Indoor Air Pollution?

Pritisha Sarkar, Kushalava reddy Jala, Mousumi Saha

TL;DR

The paper addresses indoor air pollution risk assessment by identifying pollutant sources linked to specific indoor activities using explainable AI. It combines FLOW-based sensing of $VOC$, $NO_2$, and PM metrics with clustering and interpretable models (LIME/SHAP) to attribute pollutants to activities and deliver personalized 24-hour exposure estimates. Key contributions include comprehensive multi-scenario data collection, cluster-based pollutant source identification, interpretable explanations, and a 24-hour exposure calculation framework that weights clusters by potency. This approach enables targeted, personalized interventions and has potential to support real-time alerts and behavior changes to improve indoor health outcomes.

Abstract

Acknowledging the effects of outdoor air pollution, the literature inadequately addresses indoor air pollution's impacts. Despite daily health risks, existing research primarily focused on monitoring, lacking accuracy in pinpointing indoor pollution sources. In our research work, we thoroughly investigated the influence of indoor activities on pollution levels. A survey of 143 participants revealed limited awareness of indoor air pollution. Leveraging 65 days of diverse data encompassing activities like incense stick usage, indoor smoking, inadequately ventilated cooking, excessive AC usage, and accidental paper burning, we developed a comprehensive monitoring system. We identify pollutant sources and effects with high precision through clustering analysis and interpretability models (LIME and SHAP). Our method integrates Decision Trees, Random Forest, Naive Bayes, and SVM models, excelling at 99.8% accuracy with Decision Trees. Continuous 24-hour data allows personalized assessments for targeted pollution reduction strategies, achieving 91% accuracy in predicting activities and pollution exposure.

Can Explainable AI Assess Personalized Health Risks from Indoor Air Pollution?

TL;DR

The paper addresses indoor air pollution risk assessment by identifying pollutant sources linked to specific indoor activities using explainable AI. It combines FLOW-based sensing of , , and PM metrics with clustering and interpretable models (LIME/SHAP) to attribute pollutants to activities and deliver personalized 24-hour exposure estimates. Key contributions include comprehensive multi-scenario data collection, cluster-based pollutant source identification, interpretable explanations, and a 24-hour exposure calculation framework that weights clusters by potency. This approach enables targeted, personalized interventions and has potential to support real-time alerts and behavior changes to improve indoor health outcomes.

Abstract

Acknowledging the effects of outdoor air pollution, the literature inadequately addresses indoor air pollution's impacts. Despite daily health risks, existing research primarily focused on monitoring, lacking accuracy in pinpointing indoor pollution sources. In our research work, we thoroughly investigated the influence of indoor activities on pollution levels. A survey of 143 participants revealed limited awareness of indoor air pollution. Leveraging 65 days of diverse data encompassing activities like incense stick usage, indoor smoking, inadequately ventilated cooking, excessive AC usage, and accidental paper burning, we developed a comprehensive monitoring system. We identify pollutant sources and effects with high precision through clustering analysis and interpretability models (LIME and SHAP). Our method integrates Decision Trees, Random Forest, Naive Bayes, and SVM models, excelling at 99.8% accuracy with Decision Trees. Continuous 24-hour data allows personalized assessments for targeted pollution reduction strategies, achieving 91% accuracy in predicting activities and pollution exposure.
Paper Structure (22 sections, 5 figures, 5 tables, 1 algorithm)

This paper contains 22 sections, 5 figures, 5 tables, 1 algorithm.

Table of Contents

  1. Introduction
  2. Problem statement
  3. Contribution
  4. Article organization
  5. Literature survey
  6. Health implications for indoor air quality
  7. Indoor air quality sensing
  8. System overview
  9. Data description
  10. Data distribution
  11. Methodology
  12. Identify the major pollutants from different indoor pollutant activities
  13. Clustering of indoor air pollutants
  14. Data modeling
  15. Data interpretation
  16. ...and 7 more sections

Figures (5)

  • Figure 1: System framework for indoor exposure, health risk assessment, activity prediction, with data collection, analysis, visualization, interpretation of our research work
  • Figure 2: A. Illustration depicting the interface of a Flow device and B. showcasing its components and functionalities. C. Instances of Data Acquisition Scenarios, like cooking, AC usage.
  • Figure 3: Workflow to calculate individual's indoor air pollutant intake
  • Figure 4: LIME model outcomes presenting the distinctive characteristics of Cluster 0, Cluster 1, and Cluster 2 in three separate subplots
  • Figure 5: Prediction of 24-hour indoor air pollution activities, with A representing the first activity, B as the second, C as the third, and D as the fourth activity of an individual.