Start Your Search

M. Wasdell



Author of

  • +

    OS30 - Enabling High Value Analytics (ID 44)

    • Event: e-Health 2018 Virtual Meeting
    • Type: Oral Session
    • Track: Clinical Delivery
    • Presentations: 1
    • +

      OS30.02 - Using Advanced Analytics to Predict Clinical Aggression in Mental Health (ID 368)

      M. Wasdell, Ontario Shores Centre for Mental Health Sciences; Whitby/CA

      • Abstract
      • Slides

      Purpose/Objectives: Leveraging the advanced analytics capabilities inherent in a fully integrated EHR, the hospital is piloting an analytics solution with the goal of improving capabilities to predict and clearly identify clinical aggression (CA) and enhance staff capacity to respond effectively to risks of future CA from a diverse patient cohort. Objectives: -Reducing incidents of CA, harm to others, and restraint/seclusion use while increasing therapeutic engagement -Increase confidence and competence of staff in identifying and managing the antecedents of CA -Improve safety for both clinicians and patients

      Methodology/Approach: A number of risk assessment tools for imminent aggression and discrete data are captured in the EHR; however, they have limited predictive power with respect to indicating when a patient could become aggressive. Additionally, the predictive utility of these scales are in the very near term; and the ideal predictive model would be able to predict CA risk days prior to its occurrence to have time to implement robust interventions. To address this, we have partnered with an expert to create a neural network predictive model that will leverage data found within unstructured and structured data sources in the EHR. If the model is successful at predicting CA, the implications for next steps are vast, including ability of clinicians to be alerted to become proactive to improve patient outcomes. Validation of the predictive model was completed over 2 months with a pilot focusing on the predictive models ability to accurately predict CA incidents with current data over a four month period. Pre-implementation data will be utilized to compare prediction performance rates with implementation groups. A comparison of expected verses actual number of CA incidents will be conducted utilizing reports which identify the total number of CA incidents against the expected number of CA incidents. Control charts reviewing trends for the CA incidents, harm to others, and restraint & seclusion use (incidents and duration) will be utilized to evaluate. A chart audit of CA occurances will be conducted to identify what the care planning intervention.

      Finding/Results: Implementation of the predictive model is still in progress. An evaluation framework has been developed and the predictive model will be evaluated along the following domains: -Model Performance- How accurate is the model in predicting CA? -Outcomes-Are there any improvements in key outcome and balancing indicators: Incidents of CA, harm to others (incident data – harm to staff or patients), and restraint & seclusion use All findings from the pilot will be presented at the 2018 eHealth conference.

      Conclusion/Implications/Recommendations: The implications of creating a predictive tool for CA are vast. This approach will allow staff to identify dynamic risk factors that contribute to the risk for CA and subsequently facilitate proactive refinement of care plans to mitigate or address the risk and potentially reducing aggressive incidents and restraint and seclusion use. Moreover, from a risk management and operational perspective, implementation of a predictive alerting tool for CA has the potential to positively impact patient incidents as well as staff injury rates related to patient aggression, and the subsequent sick time and costs to the healthcare system associated with these events.

      140 Character Summary: This predictive solution has the potential to reduce clinical aggression and restraint/seclusion use which would improve patient outcomes and staff safety.

      Only Active Members that have purchased this event or have registered via an access code will be able to view this content. To view this presentation, please login or select "Add to Cart" and proceed to checkout.