Prediction method for diagnosing hepatitis patient by interpretable machine learning model
The invention discloses an interpretable machine learning prediction method for diagnosing a hepatitis patient according to a blood detection result. The interpretable machine learning prediction method is characterized by mainly comprising the following steps: acquiring the blood detection result of the hepatitis patient and a hepatitis diagnosis condition; processing the missing value, and obtaining 540 positive samples and equal number of negative samples by using a data balancing strategy; performing prediction by using a black box model random forest, a support vector machine and AdaBoost; processing the model by using Bayesian optimization and grid optimization algorithms; the model with the optimal precision is selected as a final prediction model, and a prediction result is output; five evaluation indexes including AUC, accuracy, precision, F1-score and recall rate are used for measuring the model; sHAP is used for globally explaining the selected model, and LIME is used for locally explaining a prediction result. According to the method, invasive detection is not needed, whether a patient suffers from hepatitis C or not can be diagnosed through non-invasive blood detection, meanwhile, interpretability is achieved, and compared with the most advanced method, the method has better recognition performance and the prediction process is more transparent..
Medienart: |
Patent |
---|
Erscheinungsjahr: |
2023 |
---|---|
Erschienen: |
2023 |
Enthalten in: |
Europäisches Patentamt - (2023) vom: 15. Dez. Zur Gesamtaufnahme - year:2023 |
---|
Sprache: |
Englisch |
---|
Beteiligte Personen: |
FAN YONGXIAN [VerfasserIn] |
---|
Links: |
Volltext [kostenfrei] |
---|
Anmerkungen: |
Source: www.epo.org (no modifications made), First posted: 2023-12-15, Last update posted on www.tib.eu: 2024-03-13, Last updated: 2024-03-22 |
---|
Patentnummer: |
CN117238479 |
---|
Förderinstitution / Projekttitel: |
|
---|
PPN (Katalog-ID): |
EPA000651125 |
---|
LEADER | 01000caa a22002652 4500 | ||
---|---|---|---|
001 | EPA000651125 | ||
003 | DE-627 | ||
005 | 20240322142755.0 | ||
007 | cr uuu---uuuuu | ||
008 | 240315s2023 xx |||||o 00| ||eng c | ||
035 | |a (DE-627)EPA000651125 | ||
035 | |a (EPA)CN117238479 | ||
035 | |a (EPA)89090027 | ||
040 | |a DE-627 |b ger |c DE-627 |e rakwb | ||
041 | |a eng | ||
100 | 0 | |a FAN YONGXIAN |e verfasserin |4 aut | |
245 | 1 | 0 | |a Prediction method for diagnosing hepatitis patient by interpretable machine learning model |
264 | 1 | |c 2023 | |
336 | |a Text |b txt |2 rdacontent | ||
337 | |a Computermedien |b c |2 rdamedia | ||
338 | |a Online-Ressource |b cr |2 rdacarrier | ||
500 | |a Source: www.epo.org (no modifications made), First posted: 2023-12-15, Last update posted on www.tib.eu: 2024-03-13, Last updated: 2024-03-22 | ||
520 | |a The invention discloses an interpretable machine learning prediction method for diagnosing a hepatitis patient according to a blood detection result. The interpretable machine learning prediction method is characterized by mainly comprising the following steps: acquiring the blood detection result of the hepatitis patient and a hepatitis diagnosis condition; processing the missing value, and obtaining 540 positive samples and equal number of negative samples by using a data balancing strategy; performing prediction by using a black box model random forest, a support vector machine and AdaBoost; processing the model by using Bayesian optimization and grid optimization algorithms; the model with the optimal precision is selected as a final prediction model, and a prediction result is output; five evaluation indexes including AUC, accuracy, precision, F1-score and recall rate are used for measuring the model; sHAP is used for globally explaining the selected model, and LIME is used for locally explaining a prediction result. According to the method, invasive detection is not needed, whether a patient suffers from hepatitis C or not can be diagnosed through non-invasive blood detection, meanwhile, interpretability is achieved, and compared with the most advanced method, the method has better recognition performance and the prediction process is more transparent. | ||
650 | 4 | |a G06N: Computer systems based on specific computational models | |
650 | 4 | |a G16H: Healthcare informatics, i.e. information and communication technology [ict] specially adapted for the handling or processing of medical or healthcare data | |
650 | 4 | |a inf | |
650 | 4 | |a G06F: Electric digital data processing (computer systems based on specific computational models g06n) | |
650 | 4 | |a 615 | |
700 | 0 | |a LU QIANQIAN |4 aut | |
700 | 0 | |a SUN GUICONG |4 aut | |
700 | 0 | |a LIU MENG |4 aut | |
700 | 0 | |a PAN YINGJIE |4 aut | |
700 | 0 | |a ZHENG MENGXIN |4 aut | |
700 | 0 | |a WANG CHEN |4 aut | |
700 | 0 | |a LI XUEPING |4 aut | |
700 | 0 | |a GUO ZHI |4 aut | |
773 | 0 | 8 | |i Enthalten in |t Europäisches Patentamt |g (2023) vom: 15. Dez. |
773 | 1 | 8 | |g year:2023 |g day:15 |g month:12 |
856 | 4 | 0 | |u https://worldwide.espacenet.com/patent/search/family/89090027/publication/CN117238479A1?q=CN117238479 |z kostenfrei |3 Volltext |
912 | |a GBV_EPA | ||
951 | |a AR | ||
952 | |j 2023 |b 15 |c 12 |