Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Toward an Integrated Decision Making Framework for Optimized Stroke Diagnosis with DSA and Treatment under Uncertainty

Nur Ahmad Khatim, Ahmad Azmul Asmar Irfan, Amaliya Mata'ul Hayah, Mansur M. Arief

TL;DR

This work formulates a POMDP-based framework for integrated stroke diagnosis and treatment under uncertainty, combining CT imaging, Siriraj scores, and DSA within an online DESPOT solver. The model represents stroke states as $IsAne$, $IsAVM$, and $IsOcc$, with actions including observation and treatment options, and uses a belief state $b_t$ updated by Bayes’ rule in the presence of noisy observations. Experiments on $K=10{,}000$ simulated cases compare DESPOT against random and expert policies, showing DESPOT achieves near-expert performance in terms of mean cumulative discounted reward and competitive time-to-treatment, while balancing costs and invasiveness of procedures. The study demonstrates the practical value of uncertainty-aware planning for stroke care and provides open-source code for benchmarking and extension, highlighting both opportunities and limits of current data and modeling assumptions.

Abstract

This study addresses the challenge of stroke diagnosis and treatment under uncertainty, a critical issue given the rapid progression and severe consequences of stroke conditions such as aneurysms, arteriovenous malformations (AVM), and occlusions. Current diagnostic methods, including Digital Subtraction Angiography (DSA), face limitations due to high costs and its invasive nature. To overcome these challenges, we propose a novel approach using a Partially Observable Markov Decision Process (POMDP) framework. Our model integrates advanced diagnostic tools and treatment approaches with a decision-making algorithm that accounts for the inherent uncertainties in stroke diagnosis. Our approach combines noisy observations from CT scans, Siriraj scores, and DSA reports to inform the subsequent treatment options. We utilize the online solver DESPOT, which employs tree-search methods and particle filters, to simulate potential future scenarios and guide our strategies. The results indicate that our POMDP framework balances diagnostic and treatment objectives, striking a tradeoff between the need for precise stroke identification via invasive procedures like DSA and the constraints of limited healthcare resources that necessitate more cost-effective strategies, such as in-hospital or at-home observation, by relying only relying on simulation rollouts and not imposing any prior knowledge. Our study offers a significant contribution by presenting a systematic framework that optimally integrates diagnostic and treatment processes for stroke and accounting for various uncertainties, thereby improving care and outcomes in stroke management.

Toward an Integrated Decision Making Framework for Optimized Stroke Diagnosis with DSA and Treatment under Uncertainty

TL;DR

This work formulates a POMDP-based framework for integrated stroke diagnosis and treatment under uncertainty, combining CT imaging, Siriraj scores, and DSA within an online DESPOT solver. The model represents stroke states as , , and , with actions including observation and treatment options, and uses a belief state updated by Bayes’ rule in the presence of noisy observations. Experiments on simulated cases compare DESPOT against random and expert policies, showing DESPOT achieves near-expert performance in terms of mean cumulative discounted reward and competitive time-to-treatment, while balancing costs and invasiveness of procedures. The study demonstrates the practical value of uncertainty-aware planning for stroke care and provides open-source code for benchmarking and extension, highlighting both opportunities and limits of current data and modeling assumptions.

Abstract

This study addresses the challenge of stroke diagnosis and treatment under uncertainty, a critical issue given the rapid progression and severe consequences of stroke conditions such as aneurysms, arteriovenous malformations (AVM), and occlusions. Current diagnostic methods, including Digital Subtraction Angiography (DSA), face limitations due to high costs and its invasive nature. To overcome these challenges, we propose a novel approach using a Partially Observable Markov Decision Process (POMDP) framework. Our model integrates advanced diagnostic tools and treatment approaches with a decision-making algorithm that accounts for the inherent uncertainties in stroke diagnosis. Our approach combines noisy observations from CT scans, Siriraj scores, and DSA reports to inform the subsequent treatment options. We utilize the online solver DESPOT, which employs tree-search methods and particle filters, to simulate potential future scenarios and guide our strategies. The results indicate that our POMDP framework balances diagnostic and treatment objectives, striking a tradeoff between the need for precise stroke identification via invasive procedures like DSA and the constraints of limited healthcare resources that necessitate more cost-effective strategies, such as in-hospital or at-home observation, by relying only relying on simulation rollouts and not imposing any prior knowledge. Our study offers a significant contribution by presenting a systematic framework that optimally integrates diagnostic and treatment processes for stroke and accounting for various uncertainties, thereby improving care and outcomes in stroke management.
Paper Structure (14 sections, 3 equations, 6 figures, 3 tables)

This paper contains 14 sections, 3 equations, 6 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Stroke Diagnosis and Treatment
  4. Partially Observable Markov Decision Process (POMDP)
  5. POMDP for Medical Decision Making
  6. Methodology
  7. Transition Model
  8. Observation Model
  9. Belief Model
  10. Experiment Setting
  11. Discussion
  12. Performance Benchmarks
  13. Sampled Actions for Mild and Severe Cases
  14. Conclusion

Figures (6)

  • Figure 1: Illustration of simulated policy rollouts in tree search
  • Figure 2: The normalized histograms of discounted reward
  • Figure 3: The normalized histograms of time-to-treatment
  • Figure 4: The normalized discounted reward for patient with/without stroke
  • Figure 5: Sampled actions from all policies for a mild case ($IsAne: TRUE, IsAVM: FALSE, IsOcc: FALSE$)
  • ...and 1 more figures