MLD: Build-up Computed Assisted History Taking, Physical Examination and Diagnosis System of Emergency Patient Through Machine Learning (II)
Study Details
Study Description
Brief Summary
In emergency department（ED）, physicians need to complete patient evaluation and management in a short time, which required different history taking, and physical examination skill in healthcare system.
Natural language processing（NLP） became easily accessible after the development of machine learning（ML）. Besides, electronic medical record（EMR） had been widely applied in healthcare systems. There are more and more tools try to capture certain information from the EMR help clinical workers handle increasing patient data and improving patient care.
However, to err is human. Physicians might omit some important signs or symptoms, or forget to write it down in the record especially in a busy emergency room. It will lead to an unfavorable outcome when there were medical legal issue or national health insurance review. The condition could be limited by a EMR supporting system. The quality of care will also improve.
The investigators are planning to analyze EMR of emergency room by NLP and machine learning. To establish the linkage between triage data, chief complaint, past history, present illness and physical examination. The investigators will try to predict the tentative diagnosis and patient disposition after the relationship being found. Thereafter, the investigators could try to predict the key element of history taking and physical examination of the patient and inform the physician when the miss happened. The investigators hope the system may improve the quality of medical recording and patient care.
| Condition or Disease | Intervention/Treatment | Phase |
|---|---|---|
|
N/A |
Study Design
Arms and Interventions
| Arm | Intervention/Treatment |
|---|---|
| No Intervention: Control
|
|
| Experimental: Experimental
|
Diagnostic Test: Artificial intelligence
After the patients under triage classification to which randomly allocates in two groups. The group with AI intervention and the other without AI intervention.
|
Outcome Measures
Primary Outcome Measures
- Senior doctor appraisal [24 hours]
Senior doctor appraisal which measured by an established questionnaire. Senior doctor will fill an expert-verified clinical note quality evaluation questionnaire after junior doctor finished patient interview and clinical note recording. The questionnaire is designed to use 5 points likert scale and higher scores mean a better outcome.
Secondary Outcome Measures
- Accuracy of diagnosis prediction [patient discharge from ED, up to 1 week]
The percentage of predicted diagnosis match the final diagnosis.
- Rationality of diagnosis prediction [24 hours]
Senior doctors will assess rationality of predicted diagnosis.
Eligibility Criteria
Criteria
Inclusion Criteria:
-
Over twenty years old
-
Non-traumatic patient
Exclusion Criteria:
-
Excluding the patients for administration reasons (issuing a medical certificate)
-
Excluding the patients for non-emergency reasons like simply acupuncture, virus screening and prescription for medication.
-
Excluding Patients who allocated to critical care station
Contacts and Locations
Locations
No locations specified.Sponsors and Collaborators
- National Taiwan University Hospital
Investigators
- Study Chair: Huang, Dr., National Taiwan University Hospital
Study Documents (Full-Text)
None provided.More Information
Publications
None provided.- 202110012RIND