Unique ID issued by UMIN | UMIN000036064 |
---|---|
Receipt number | R000041083 |
Scientific Title | Comparison between manual analysis of medical records and analysis using machine learning |
Date of disclosure of the study information | 2019/04/30 |
Last modified on | 2024/09/03 09:42:49 |
Comparison between manual analysis of medical records and analysis using machine learning
Comparison between manual analysis of medical records and analysis using machine learning
Comparison between manual analysis of medical records and analysis using machine learning
Comparison between manual analysis of medical records and analysis using machine learning
Japan |
Headache
Medicine in general | Emergency medicine |
Others
NO
Symbol sign determination is essential for constructing automatic diagnostic system using Bayes' theorem. Text mining technology has made it possible to process enormous amounts of information in a short time. Mechanical sign judgment of a medical record which is a natural language is said to be difficult due to the structure of Japanese. For the free description part in the medical record, compare the accuracy of the sign judgment of the symptom extracted comprehensively manually and the sign judgment extracted using the machine learning.
Efficacy
Percentage of combinations that generated cross table
Sensitivity, specificity and likelihood ratio of each symptom
Observational
20 | years-old | <= |
Not applicable |
Male and Female
Patients who visited the Tokyo Metropolitan Tama General Medical Center emergency outpatient for 2 months from May 1, 2014, with a headache complaint
none
270
1st name | Daiki |
Middle name | |
Last name | Yokokawa |
Chiba University Hospital
Department of General Medicine
2608677
1-8-1, Inohana, Chuo-ku, Chiba city
0432227171
dyokokawa6@chiba-u.jp
1st name | Daiki |
Middle name | |
Last name | Yokokawa |
Chiba University Hospital
Department of General Medicine
2608677
1-8-1, Inohana, Chuo-ku, Chiba city
0432227171
dyokokawa6@chiba-u.jp
Chiba University
None
Other
Chiba University Hospital
1-8-1, Inohana, Chuo-ku, Chiba city
0432227171
dyokokawa6@chiba-u.jp
NO
2019 | Year | 04 | Month | 30 | Day |
https://www.cureus.com/articles/231002-a-cross-sectional-study-on-whether-comprehensively-gathering-
Published
https://www.cureus.com/articles/231002-a-cross-sectional-study-on-whether-comprehensively-gathering-
270
Probability functions for the appearance of new unique keywords were modeled, and theoretical values were calculated. We extracted 623 unique keywords, 26 diagnoses, and 6,904 annotated keywords. Likelihood ratios could be calculated only for 276 combinations (1.70%), of which 24 (0.15%) exhibited significant differences. The power function+constant was the best fit for new unique keywords.
2024 | Year | 09 | Month | 03 | Day |
2024 | Year | 06 | Month | 04 | Day |
This was a single-center study. We retrospectively extracted the MRs of patients aged >=16 years whose chief complaint was experiencing headaches and who visited the emergency room (ER) at the Tokyo Metropolitan Tama Medical Center between May 1 and June 30, 2014. Approximately 150 patients experiencing headaches visit the hospital every month. The number of patients' records that could be annotated by reviewing all MRs was approximately 300.
Patients were included in the study if they visited the hospital on their own or by ambulance. In Japan, pediatricians treat patients under 16 years of age. The hospital under study does not have a pediatric department; therefore, patients under 16 years of age were excluded. Similarly, if patients were in a severe or critical condition, having experienced a stroke or shock based on vital signs and symptoms noted in the ER or ambulance, they were transferred to a critical emergency center, which became responsible for tertiary care, and thus were excluded from the study.
NA
Diagnosis names and number of cases.
Frequency of unique keywords, annotated keywords, and calculated combinations
Operational characteristics
Completed
2019 | Year | 03 | Month | 01 | Day |
2019 | Year | 04 | Month | 22 | Day |
2019 | Year | 04 | Month | 22 | Day |
2020 | Year | 08 | Month | 31 | Day |
Two physicians extracted all symptoms from the free entry in the medical record and signed it. Diagnosis was done using international headache classification 2nd edition, and a cross table of symptoms and diagnosis was prepared. Calculate the proportion of combinations that could produce a cross table and the sensitivity, specificity and likelihood ratio of each symptom.
Calculate the same indicator using machine learning, and compare them respectively.
2019 | Year | 03 | Month | 01 | Day |
2024 | Year | 09 | Month | 03 | Day |
Value
https://center6.umin.ac.jp/cgi-open-bin/ctr_e/ctr_view.cgi?recptno=R000041083