AFGuide System to Support Personalized Management of Atrial Fibrillation

Transcription

1 AFGuide System to Support Personalized Management of Atrial Fibrillation Martin Michalowski; Wojtek Michalowski; Dympna O Sullivan; Szymon Wilk; Marc Carrier

2 Outline Motivation Existing Tools AFGuide Discussion

3 Motivation Atrial fibrillation (AF) constitutes a serious public health problem Aging population [Chugh et al. 2014] Oral anticoagulant (OAC) therapy is sub-optimal [Cotte et al. 2014] Vitamin K antagonists (VKAs), e.g. Warfarin Newly developed direct OACs (DOACs) are: [Giugliano et al. 2013] Faster More predictable and sustainable anticoagulation Similar effectiveness as VKAs in stroke prevention Lower bleeding and risk of death from CV causes

4 What s The Problem? OAC management is suboptimal Only 10% of AF patients with first acute stroke were therapeutically anticoagulated at the time of admission [Gladstone et al., 2009] A significant portion of primary care patients not treated according to AF CPGs [Valentinis et al. 2014] Knowledge gap present for primary care physicians (PCPs) Optimal OAC therapy can prevent approximately 80% of AF-related strokes

5 Barriers to Knowledge Uptake [Murray et al. 2011] 1. Lack of up-to-date knowledge about new therapies 2. Difficulty using clinical practice guidelines (CPGs) for multimorbidity 3. Lowered compliance due to lack of patients engagement AFGuide A clinical decision support system (CDSS) to educate and support PCPs in developing optimal OAC therapy to improve patient management

6 Existing Tools

7 AFGuide* *Proposed system being implemented

8 AFGuide Medical & Patient Context Executable CPG Canadian Cardiovascular Society s (CCS) CPG for AF Potential adverse events when managing a multimorbid patient Represented as an actionable graph based on SDA* Patient Adherence-to-Therapy and Preference Models Create baseline models using Generalized Regression with Intensities of Preference (GRIP) method [O Sullivan et al. 2014] Represented as additive value function Capture and operationalize patient preferences [Michalowski et al. 2015] Advanced evaluation of derived OAC therapies

9 AFGuide

10 AFGuide - Stroke and Bleeding Risk Assessor Risk-related operators represented as a First-Order Logic (FOL) theory Compute stroke and bleeding risk scores based on patient data Stroke risk (CHA 2 DS 2 -VASc) Bleeding risk (HAS-BLED) Takes into account patient context for comprehensive assessment Used in therapy development

11 AFGuide

12 AFGuide - Therapy Generator and Explainer *Generator* Derives feasible OAC therapies using FOL [Wilk et al., 2016] Theorem proving and model finding techniques Mitigates any adverse interactions (multimorbidity) Ranks recommended OAC therapies according to their confidence levels Stroke and bleeding risk assessment Scoring functions associated with adherence-to-therapy and preference models PCP can select any therapy from the list Revise the patient context *Explainer* On-request justification for each therapy in the ranked list Reviews from the Cochrane Database of Systematic Reviews [O Sullivan et al. 2010]

13 AFGuide

14 AFGuide - Model Manager and Learner *Management* Automatically selects the adherence-to-therapy and preference models for new patient *Learning* During PCP-patient encounter Observes interactions and therapy choices Customizes baseline adherence-to-therapy and preference models for a given patient using feedback loop Transforms population-based baseline models into customized patient-specific ones

15 AFGuide Tools for Implementation

16 Discussion - Goals Narrow the knowledge gap among PCPs w.r.t. optimal OAC therapy for patients with AF 1. Decision support for PCPs Address the complexities of evidence-based OAC therapy development 2. Derive an evidence-based OAC therapy Patient-specific Multimorbidity Patient Preferences PCP education

17 Discussion Current Status Designed AFGuide Developed: executable CPG, FOL models, therapy explainer and generator Under development: model manager and learner, stroke and bleeding risk assessor Ongoing: Integration of all components Ongoing: Working with clinical partners to design the evaluation studies

18 Questions? Thank you! MET Research:

19 References Chugh, S. S.; Havmoeller, R.; Narayanan, K.; Singh, D.; Rienstra, M.; Benjamin, E. J.; Gillum, R. F.; Kim, Y.-H.; McAnulty, J. H.; Zheng, Z.-J.; Forouzanfar, M. H.; Naghavi, M.; Mensah, G. A.; Ezzati, M.; and Murray, C. J Worldwide epidemiology of atrial fibrillation. Circulation 129(8): Cotte, F. E.; Benhaddi, H.; Duprat-Lomon, I.; Doble, A.; Marchant, N.; Letierce, A.; and Huguet, M Vitamin k antagonist treatment in patients with atrial fibrillation and time in therapeutic range in four European countries. Clin Ther 36(9): Giugliano, R. P.; Ruff, C. T.; Braunwald, E.; Murphy, S. A.; Wiviott, S. D.; Halperin, J. L.; Waldo, A. L.; Ezekowitz, M. D.; Weitz, J. I.; Spinar, J.; Ruzyllo, W.; Ruda, M.; Ko- retsune, Y.; Betcher, J.; Shi, M.; Grip, L. T.; Patel, S. P.; Pa- tel, I.; Hanyok, J. J.; Mercuri, M.; and Antman, E. M Edoxaban versus warfarin in patients with atrial fibrillation. N Engl J Med 369(22): Gladstone, D. J.; Bui, E.; Fang, J.; Laupacis, A.; Lindsay, M. P.; Tu, J. V.; Silver, F. L.; and Kapral, M. K Potentially preventable strokes in high-risk patients with atrial fibrillation who are not adequately anticoagulated. Stroke 40(1): Valentinis, A.; Ivers, N.; Bhatia, S.; Meshkat, N.; Leblanc, K.; Ha, A.; and Morra, D Atrial fibrillation anticoag- ulation care in a large urban family medicine practice. Can Fam Physician 60(3):e173 9 Murray, S.; Lazure, P.; Pullen, C.; Maltais, P.; and Dorian, P Atrial fibrillation care: challenges in clinical practice and educational needs assessment. Can J Cardiol 27(1):98 104

20 References Michalowski, M.; Wilk, S.; Rosu, D.; Kezadri, M.; Michalowski, W.; and Carrier, M Expanding a firstorder logic mitigation framework to handle multimorbid patient preferences. AMIA Annu Symp Proc O Sullivan, D.; Wilk, S.; Michalowski, W.; Slowinski, R.; Thomas, R.; Kadzinski, M.; and Farion, K. 2014b. Learning the preferences of physicians for the organization of result lists of medical evidence articles. Methods Inf Med 53(5): Wilk S.; Michalowski, M.; Michalowski, W.; Rosu, D.; Carrier, M.; Kezadri-Hamiaz, M Comprehensive Mitigation Framework for Concurrent Application of Multiple Clinical Practice Guidelines. Journal of Biomedical Informatics. (In press) O Sullivan, D. M.; Wilk, S.; Michalowski, W. J.; and Farion, K. J Automatic indexing and retrieval of encounter- specific evidence for point-of-care support. J Biomed Inform 43(4): U.S. Department of Health & Human Services Sys- tem usability scale (SUS). and-tools/methods/system-usability-scale.html

21 Designed Evaluation Two-phase evaluation study 1. Usability study involving PCPs Interact with AFGuide when deriving a therapy using clinical vignettes System Usability Scale [U.S. Department of Health & Human Services 2016] 2. Clinical validation In collaboration with selected Canadian hospitals