Presentation is loading. Please wait.

Presentation is loading. Please wait.

Decaying Relevance of Clinical Data when Predicting Future Decisions

Similar presentations


Presentation on theme: "Decaying Relevance of Clinical Data when Predicting Future Decisions"— Presentation transcript:

1 Decaying Relevance of Clinical Data when Predicting Future Decisions
Presenter: Muthu Alagappan, MD Jonathan Chen, MD, PhD Mary K Goldstein, MD, MS Steven M Asch MD, MPH Russ B Altman MD, PhD March 30th, 2017 Joint Summits on Translational Science General Medicine Disciplines, Department of Medicine, Stanford University

2 Disclosures I do not have any relevant financial relationships or conflicts of interest

3 Objectives To develop a clinical order recommender system
To determine how varying historical training data affects prediction To estimate the “decay rate” (i.e. relevance) of clinical data

4

5

6 Randomized Clinical Trials (RCTs)
Knowledge Scientific inquiry Randomized Clinical Trials (RCTs) Clinical Orders Patient Care Expert Opinion Anecdotal Experience

7 Clinical Decision Support - Order Sets

8 Clinical Decision Support - A bottom-up approach
Collaborative filter Doctor Collective Medical Record

9 Data Description

10 Patient Timeline Model

11 Example: Chest Pain (ICD9: 786.5)
Top orders following an admission diagnosis of “Chest Pain” Co-occurrences within 24 hours, ranked by Odds Ratio (OR)

12 Internal Validation Use initial data as query
Predict subsequent orders Compare against next actual 24 hour orders Identify real order set use

13 Prediction Accuracy

14 Add more data? “…invariably, simple models and a lot of data trump more elaborate models based on less data.” “…use available large-scale data rather than hoping for annotated data that isn’t available.”

15 Is More Data Better? Prediction is difficult
Clinical knowledge evolves quickly Recent data > More data

16 Decaying Relevance of Clinical Data

17 Objectives To develop a clinical order recommender system
To determine how varying historical training data affects prediction To estimate the “decay rate” (i.e. relevance) of clinical date

18 Acknowledgements Mentor Jonathan Chen, MD, PhD


Download ppt "Decaying Relevance of Clinical Data when Predicting Future Decisions"

Similar presentations


Ads by Google