Application of a Machine Learning Algorithm in Prediction of Abusive Head Trauma in Children
Priyanka Jadhav1, Timothy Sears2, Gretchen Floan3
1University of California San Diego School of Medicine, 9500 Gilman Dr, La Jolla, CA, 92093, USA.
Insights
Machine learning accurately detects abusive head trauma (AHT) in children using physician notes and demographics. This algorithm improves early AHT identification, aiding timely intervention and improving patient outcomes.
Area of Science:
- Medical Informatics
- Pediatric Emergency Medicine
- Machine Learning in Healthcare
Background:
- Abusive head trauma (AHT) is a serious concern in pediatric care.
- Timely detection of AHT is crucial for intervention and improving outcomes.
- Current diagnostic methods can be time-consuming and may miss subtle indicators.
Purpose of the Study:
- To develop and evaluate a machine learning algorithm for early detection of potential abusive head trauma (AHT).
- To utilize free-text physician notes and demographic data for AHT identification.
- To assess the algorithm's performance in distinguishing AHT cases from other head injuries.
Main Methods:
- Collected first free-text physician notes and demographic data for children under 5.
- Compared AHT cases (diagnosed by Child Protective Team) with a control group (head/neck injury).
- Developed composite scores (differential, sentiment, subjectivity) and trained a Random Forest algorithm with demographic data.
Main Results:
- The Random Forest model combining composite scores and demographics achieved an 84% accuracy.
- Area under the curve (AUC) improved from 0.68 to 0.78 when incorporating demographic data.
- Key predictors included composite score, sentiment, age, and subjectivity; subjectivity trended higher in AHT cases.
Conclusions:
- Machine learning effectively identifies patterns in clinical notes and demographics indicative of AHT.
- This approach offers a promising tool for timely and accurate AHT detection in children.
- Further integration of such algorithms can enhance pediatric patient safety and care.
Purpose:
We explored the application of a machine learning algorithm for the timely detection of potential abusive head trauma (AHT) using the first free-text note of an encounter and demographic information.
Methods:
First free-text physician notes and demographic information were collected for children under 5 years of age at a Level 1 Trauma Center. The control group, which included patients with head/neck injury, was compared to those with AHT diagnosed by the Child Protective Team. Differential scores accounted for words overrepresented in AHT patient vs. control notes. Sentiment scores were reflective of note positivity/negativity and subjectivity scores accounted for note subjectivity/objectivity. The composite scores reflected the patient's differential score modified by the subjectivity score. Composite, sentiment, and subjectivity scores combined with demographic information trained a Random Forest (RF) machine learning algorithm to predict AHT.
Results:
Final composite scores with demographic information were highly associated with AHT in a test dataset. The control group included 587 patients and the test group included 193 patients. Combining composite scores with demographic information into the RF model improved AHT classification area under the curve (AUC) from 0.68 to 0.78, with an overall accuracy of 84%. Feature importance analysis of our RF model revealed that composite score, sentiment, age, and subjectivity were the most impactful predictors of AHT. The sentiment was not significantly different between control and AHT notes (p = 0.87), while subjectivity trended higher for AHT notes (p = 0.081).
Conclusion:
We conclude that a machine learning algorithm can recognize patterns within free-text notes and demographic information that aid in AHT detection in children.
Level Of Evidence:
III.
