Clinical Decision Support Systems

Clinical decision support systems l.jpg
1 / 56
0
0
817 days ago, 227 views
PowerPoint PPT Presentation

Presentation Transcript

Slide 1

Clinical Decision Support Systems Ida Sim, MD, PhD March 12, 2002 Division of General Internal Medicine, and the Graduate Group in Biological and Medical Informatics UCSF Copyright Ida Sim, 2002. All government and state rights saved for all unique material introduced in this course through any medium, including address or print. Clinical Decision Support Systems Medical Informatics

Slide 2

IT and Quality Information innovation touted to enhance nature of care Dimensions data accessibility diagram, lab comes about, hypersensitivities; all clear procedure proficiency visit level coding, e-recommending halfway measures immunization and screening rates tolerant results Clinical Decision Support Systems Medical Informatics

Slide 3

Outline Clinical choice emotionally supportive networks (CDSS) definition strategies for thinking viability at enhancing quality Clinical research informatics foundation for clinical research frameworks for supporting clinical research Clinical Decision Support Systems Medical Informatics

Slide 4

What is a CDSS? Programming that is intended to be an immediate guide to clinical basic leadership in which the qualities of an individual patient are coordinated to a modernized clinical learning base, and patient-particular appraisals or suggestions are then displayed to the clinician and additionally the patient for a choice (Sim et al, JAMIA, 2001) Clinical Decision Support Systems Medical Informatics

Slide 5

Major Objectives Diagnostic support DxPlain, QMR Drug dosing aminoglycoside, theophylline, warfarin Preventive care updates immunizations, mammograms Disease administration diabetes, hypertension, AIDS, asthma Test requesting, medicate remedy decreasing every day CBCs in doctor's facility, sensitivity checking Utilization referral administration, center followup Clinical Decision Support Systems Medical Informatics

Slide 6

How Do CDSSs "Think"? A few frameworks utilize more than one technique administer based adhoc non-math strategy for thinking about probabilities e.g., if high WBC AND hack AND fever AND abn. CXR then probability of pneumonia is 4 out of 5 e.g., DxPlain, QMR bayesian system formal probabilistic thinking, augmentation of choice examination neural system fluffy rationale, hereditary calculations, case-based thinking, and so forth. Clinical Decision Support Systems Medical Informatics

Slide 7

Rule-Based Approaches Forward thinking (information driven) begin with information, execute material principles, check whether new conclusions trigger different guidelines, thus on utilize if meager information if high WBC AND hack AND fever AND abn. CXR => pneumonia if pneumonia => give anti-toxins, and so on. In reverse thinking (objective driven) begin with "objective administer," figure out if objective run is valid by assessing reality of every important introduce utilize if bunches of information patient with loads of discoveries and manifestations is this lupus? => are at least 4 ACR criteria fulfilled? Clinical Decision Support Systems Medical Informatics

Slide 8

library: reason: Recommend the utilization of ampicillin for pneumonia.;; clarification: If the patient has pneumonia, then propose treatment with ampicillin unless there is a penicillin sensitivity.;; catchphrases: pneumonia; penicillin; ampicillin;; references: 1. Frame Manual, adaptation 1.6. LDS Hospital, August 1989, p.81.;; MLMs and Arden Medical Logic Modules (MLMs) in Arden Syntax (a global ASTM standard linguistic structure) : help_amp_for_pneumonia - Ampicillin for Pneumonia upkeep: title: Ampicillin for Pneumonia;; filename: help_amp_for_pneumonia;; form: 1.00;; foundation: LDS Hospital;; creator: Peter Haug, M.D.; George Hripcsak, M.D.;; expert: ;; date: 1991-05-28;; approval: testing;; Clinical Decision Support Systems Medical Informatics

Slide 9

Neural Networks Example of an information driven information mining strategy Finds a non-straight relationship between information parameters and yield state Structure of system normally information, yield, and 1-2 concealed completely associated layers every association has a "weight" Clinical Decision Support Systems Medical Informatics

Slide 10

EKG discoveries Acute MI Rales No Acute MI JVD Response to TNG Neural Network for MI Diagnosis Inputs (e.g., every patient trademark in the EMR) EKG discoveries (ST rise, old Q's) rales (Yes, No) JVD (in cm) Outputs are the arrangement of conceivable results/analyze Clinical Decision Support Systems Medical Informatics

Slide 11

Training the Neural Network gets "prepared" nourish organize numerous cases of known patients and their analyses framework iteratively modifies the weights of the associations with discover the system "design" that partners sets of information factors (patients) with the right yield state (MI or not) In Baxt's MI neural system preparing set: 130 pts with MI, 120 without test set: 1070 ER patients with front mid-section torment Clinical Decision Support Systems Medical Informatics

Slide 12

Baxt's Acute MI Neural Net Evaluation comes about: predominance of MI 7% (Lancet, 1996) Results were driven by non-standard indicators rales, jugular venous extension Why isn't this neural system utilized all the more generally? "discovery" nature limits illustrative capacity and decreases acknowledgment clients need to enter the factors physically if EMRs all the more broadly accessible, these sorts of frameworks might be more common Clinical Decision Support Systems Medical Informatics

Slide 13

CDSS Methods Vast greater part of clinically-utilized CDSSs utilize govern based thinking issue of combinatorial blast of tenets Major confinements how to speak to a few information (e.g., "looks wiped out") formal, reproducible techniques for settling on clinical choices Other real constraint is wellspring of info information manual contribution of information by specialists won't work EMR can empower another time of CDSSs But parcels should be possible with ebb and flow innovation Clinical Decision Support Systems Medical Informatics

Slide 14

Outline Clinical choice emotionally supportive networks (CDSS) definition strategies for thinking viability at enhancing quality Clinical research informatics foundation for clinical research frameworks for supporting clinical research Clinical Decision Support Systems Medical Informatics

Slide 15

CDSS Effectiveness In controlled trials, just incidental humble advantage discovered (Hunt, JAMA 1998; overhauled RB Haynes 2000) analysis: 1/5 contemplates helpful medication dosing: 9/15 preventive care updates: 19/26 Few studies took a gander at patient results 6 of 14 indicated advantage Clinical Decision Support Systems Medical Informatics

Slide 16

Shortcomings of Literature Variable study quality 35% rate >8 on 10 point quality scale (mean ~6.2) later concentrates better quality Low power 5 of 8 investigations of patient result had low power Patients randomized to CDSS or not doctors treated a few patients with CDSS, and some without CDSS this would tend to … . any impacts of the CDSS Probably production predisposition Clinical Decision Support Systems Medical Informatics

Slide 17

Shortcomings of Approach (1) E.g., a hypertension treatment CDSS Is RCT best plan for deciding viability? ought to randomize MDs, \ normally low power intercession is typically more than simply the CDSS e.g., "purchase in" sessions to HTN rule basic CDSS constrained generalizability applies just to this specific CDSS joining of CDSS into existing work process regularly exceptional to study site if CDSS demonstrates no impact, standard RCT gives little understanding into why Clinical Decision Support Systems Medical Informatics

Slide 18

Shortcomings of Approach (2) How might you enhance the Hunt deliberate survey? CDSSs are extremely heterogeneous does the heterogeneity clarify any variety in advantage? Illustration: preventive care update CDSS An assistant routinely abstracts preventive care intercessions from paper outline into a database. Prior to every facility session, nurture runs the CDSS for patients coming in that day. Rule construct proposals are printed out in light of paper and connected to front of outline. Specialist orders preventive care in common way utilizing paper-based techniques Clinical Decision Support Systems Medical Informatics

Slide 19

Heterogeneity of CDSSs Hypertension treatment CDSS Clinic has an EMR. Amid patient visit, CDSS noticed that BP and pattern is too high. Checks patient's Cr, diabetes status, heart status, current meds and hypersensitivities and prescribes medicate treatment change as indicated by JNC VI rules. Presents e-medicine for MD to check. In the event that checked, request is sent specifically to drug store and pharmaceutical rundown redesigned. How to definitively portray CDSSs? target leader (MD, nurture, quiet) desperation of choice (detail result, outpatient screening) strategy for conveyance (paper, EMR, pager) drive of proposal (recommendation, prerequisite) ... Clinical Decision Support Systems Medical Informatics

Slide 20

CONTEXT Clinical choice Target tolerant setting Point of care Question introduction Workflow joining OUTPUT Action intricacy Action installed Compliance criticalness Force activity proposal Decision center Form data era CDSS Customization Update instrument Unit of investigation Clinical learning source Mode of data era INPUT Data source Data source-framework go-between System-UI System-UI OR System client/Processor/Target leader Target chief System client Processor Typology of CDSSs Clinical Decision Support Systems Medical Informatics

Slide 21

CDSS Effectiveness Summary Current information recommends CDSSs can enhance the procedure of care and maybe clinical results best at preventive care updates humble, best case scenario for medication dosing and dynamic care for the most part not accommodating for enhancing conclusion aside from with students Findings constrained by methodological issues decision of study outline lacking energy about work process segment of CDSSs Clinical Decision Support Systems Medical Informatics

Slide 22

Summary on CDSSs Intense enthusiasm for guarantee of CDSSs to enhance human services quality Evidence is equi

SPONSORS