Objective

2

To construct a clinical prediction model for perimenopausal symptoms in patients with premature ovarian insufficiency （POI） treated by

Bushen Huoxue Formula

（补肾活血方）.

Methods

2

A retrospective study was conducted to collect data of POI patients with perimenopausal symptoms who were treated by

Bushen Huoxue Formula

for six months， including age， reproductive history such as pregnancy， delivery and miscarriage history， menstrual history such as age at menarche， menstrual cycle， period and volume， marital status， age at diagnosis， course of disease， serum reproductive endocrine indicators such as follicle stimulating hormone （FSH）， luteinizing hormone （LH）， estradiol （E2）， prolactin （PRL）， progesterone （P）， testosterone （T） and anti-Müllerian hormone （AMH）， ultrasound related indicators such as peak systolic blood flow velocity （PSV） of ovarian artery and blood flow resistance index （RI）， and clinical symptoms. The patients were divided into effective and ineffective groups using the modified Kupperman score. Factors affecting the treatment outcomes of menopausal symptoms by

Bushen Huoxue Formula

in POI patients were screened out by using the least absolute shrinkage selection operator （LASSO） regression method， and were used to construct a logistic regression model and independent prediction variables. The goodness-of-fit of the model was evaluated by the Hosmer-Lemeshow test， and the area under curve （AUC） of the receiver operating characteristic （ROC）， calibration curve， and clinical decision curve analysis （DCA） were used to eva-luate the discrimination， calibration and clinical impact of the model， respectively.

Results

2

Among 204 POI patients with perimenopausal symptoms treated by

Bushen Huoxue Formula

， there were 152 cases in the effective group and 52 cases in the ineffective groups. LASSO regression analysis found seven variables including marital status， course of disease， menstrual period， FSH， LH， AMH， and ovarian vascular PSV that affected the effect. From multivariate Logistic regression analysis， we preliminarily obtained that course of disease （2.44±1.61 years）， FSH level （25-40 U/L）， AMH level （0.85±0.28 ng/ml）， PSV 8.50 （7.00， 10.38） cm/s， prolonged menstruation， and shortened menstruation were the independent variables that influenced the effect of

Bushen Huoxue Formula

on POI patients with perimenopausal symptoms. The AUC of the constructed predication model for perimenopausal symptoms in patients with POI treated by

Bushen Huoxue Formula

was 0.78. The prediction probability through DCA was 60%， that is， about 50 out of 100 patients who have received treatment can obtain better effect. When the prediction probability of the calibration curve was 30% to 85%， the prediction curve of the model was relatively close to the actual observed curve.

Conclusion

2

The constructed clinical prediction model suggests that

Bushen Huoxue Formula

is suitable for POI patients with perimenopausal symptoms who have a long course of disease， good basic ovarian function， and menstrual disorder. The prediction model can help adjusting patients' medical expectations， distinguish dominant people， and provide ideas for the precise treatment of perimenopausal symptoms in POI with

Bushen Huoxue Formula

.