| Objective:By clustering the dosages of methadone maintenance treatment patients,exploring the individual characteristics of different dosage groups,and then achieving the purpose of maintaining the therapeutic dose of methadone based on the patient's individual condition and accurately providing the desired therapeutic effect,providing data support for medication management.Methods:K-means clustering was performed on the patient's initial dose,the last dose,and the median and interquartile range of the patient's dose.The single factor comparison analysis was performed on the characteristics of different types of patients obtained by clustering.The counting data is analyzed by chi-square test,and the measurement data is analyzed by variance.Finally,the decision tree classification model is established to obtain different types of patient feature sets.Result:According to the K-means clustering results,169 patients with methadone maintenance treatment were divided into low-dose and high-dose.There were 79 patients in the low-dose group and 90 patients in the high-dose group.Two dose-separated key points were obtained,with a maintenance dose of 51 mg/d for the low-dose population and an average of 74.3 mg/d for the high-dose population.Therefore,a low dose of less than 51 mg/d,a medium dose of 51-74.3 mg/d,and a high dose of greater than 74.3 mg/d are defined.There are differences in demographics between the two types of patients.In the high-dose group,the higher-education populations in junior high school and above account for a slightly larger proportion,but the difference is small(the high-dose group has a junior high school or higher education ratio of 92.6%,The proportion of the low-dose group was 74.9%.The difference of educational level between the different dose groups was statistically significant(?2=7.139,P=0.0281<0.05).In the form of drug use,the use of simple oral or hot-sucking Compared with 65%,the proportion of injection-related drug use was 35%.It is worth noting that in the high-dose group,the proportion of drug-using methods was 44.4%,and in the low-dose group,23.9%,different doses.The difference in drug use between the groups was statistically significant(?2=4.6026,P=0.0319<0.05).Patients who received injecting drug use were more addicted than patients who received simple drug use,so the dose was higher.In terms of maintenance time,the high-dose maintenance treatment time was longer,the average duration of low-dose patients was 675.9 days,and the average of high-dose patients was 723.9 days.The difference in maintenance time between different dose groups was statistically significant.Yi(W=912,P=0.0109<0.05),indicating that the average dose of methadone in the maintenance period will last longer and the treatment effect will be better;in terms of living conditions,86% of patients with family and friends Inhabited,14% of patients were living alone,and the proportion of people living alone in high doses was larger.The number of people living alone in low doses was only 4.3%,and the number of people living alone in high doses was 22.2%.The difference in living conditions between the different dose groups was statistically significant(?2=6.5916,P=0.0102<0.05).Then the classification model is established and the data is returned.The accuracy of the decision tree model is 84.9%.Two sets of characteristics are obtained,and six rules are more effective.Rule 1: The living conditions are not solitary,and the mode of drug use is related to the injection,and the patients with work are 100% probability to be divided into high dose groups.Rule 2: If the living condition is non-living,there is a 92% probability of being divided into high dose groups.Rule 3: The living condition is non-living.The drug use method is related to the injection,and the drug friends frequently interact with each other.The time spent on the road to the clinic every day for 25 to 40 minutes is 75%.High dose group.Rule 4: The living conditions are non-living,and the way of taking drugs is related to injections,and frequent interactions with drug friends.The time spent on the way to the outpatient clinic every day is 10 to 25 minutes or 40 to 90 minutes,no work,suffering from Patients with hepatitis C have a 75% probability of being divided into high dose groups.Rule 5: The living condition is non-living,and the drug use in the past six months is simple injection,no work and drug abuse lasts for 11 to 21 years,then he has a 71% probability of being divided into high dose group.Rule 6: The living condition is non-living.In the past six months,the drug-using method is simple oral or hot-sucking.Patients who have occasionally interacted with drug users in the past month will not be classified into low-dose categories.It can be seen from the first five rules that high-dose patients are mainly related to living conditions,drug use patterns,frequency of interactions with drug users,time spent at clinics,working status,and history of hepatitis C.All the factors mentioned are living conditions with family and friends,drug use by injection and other methods,often contact with drug users,and a long history of drug abuse.Because of the degree of addiction of patients,the degree of addiction is related to the dose,and patients who are often associated with drug users and who have a long history of drug use will have higher doses during the maintenance period.Conclution:According to the baseline characteristics of methadone patients,it is possible to classify and guide the medication,and the patients are divided into high-dose group and low-dose group.Patients can be more aware of their medication category and better maintain treatment.For the manager,it is possible to pre-determine the category to which the patient belongs,and to achieve data support for medication management. |
PDF Full Text Request