Establishment method of antibody-mediated autoimmune encephalitis prognosis evaluation column map model

Abstract

The invention discloses a method for establishing an antibody-mediated autoimmune encephalitis prognosis evaluation column diagram model. The method comprises the following steps: collecting patient demographic data, clinical manifestation, vEEG, multi-mode MRI, cognition and emotion scale, blood and cerebrospinal fluid assay and follow-up data; screening independent risk factors and protection factors through single-factor and multi-factor binary Logistic regression; and establishing a column graph prediction model according to the screened factors, and evaluating the performance of the model through an ROC curve, a correction curve and DCA. The model shows abnormal mental behaviors, the epileptic seizure type FBTCS is an independent risk factor of poor prognosis, normal number of neutrophils in blood is a protection factor, marginal system atrophy is an independent risk factor of transformation of chronic epilepsy, and normal number of lymphocytes in blood is a protection factor. The method can provide quantitative and visual individualized prognosis evaluation for the antibody-mediated autoimmune encephalitis patients, and can assist in early precise diagnosis and treatment.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of neuroscience and clinical medicine, and particularly relates to a method for establishing an antibody-mediated autoimmune encephalitis prognosis evaluation nomogram model. BACKGROUND

[0002] Antibody-mediated AE is a neurological disease characterized by immune-mediated encephalitis, and its clinical manifestations are complex, including acute seizures, mental and behavioral abnormalities, and cognitive dysfunction. Various types of antibody-mediated AE have some common clinical features, but also have their own characteristics. Some AE patients can progress to chronic epilepsy, leaving cognitive dysfunction, and are accompanied by poor prognosis. In order to improve the prognosis of patients, it is necessary to study the factors affecting the prognosis of patients and intervene early. Therefore, it is of great significance to identify factors affecting early prognosis, and intervention on factors affecting early prognosis can improve the condition of patients and reduce the social burden.

[0003] At present, the assessment of AE prognosis and chronic epilepsy risk in clinical practice mainly relies on single indicators or empirical judgment, and there is a lack of systematic prediction model, making it difficult to perform early precise intervention. Previous studies have shown that clinical manifestations, electroencephalogram (EEG), magnetic resonance imaging (MRI), blood and cerebrospinal fluid test indicators are related to prognosis, but a quantifiable combined prediction system has not yet been established. Therefore, it is urgent to develop a scientific method that can integrate multi-dimensional data, quantify the risk of prognosis, and predict the outcome of chronic epilepsy, providing a basis for precise clinical diagnosis and treatment. SUMMARY

[0004] The purpose of the present application is to provide a method for establishing an antibody-mediated autoimmune encephalitis prognosis evaluation nomogram model.

[0005] A method for establishing an antibody-mediated autoimmune encephalitis prognosis evaluation nomogram model, according to the following steps:

[0006] (1) Collecting antibody-mediated autoimmune encephalitis patient data;

[0007] (2) Single factor and multi-factor binary Logistic regression analysis of the included clinical characteristics to screen independent risk factors and protective factors;

[0008] (3) Establishing a nomogram model according to the regression coefficients of the binary Logistic regression model to quantify the contribution of each predictor to the prognosis;

[0009] (4) Drawing a ROC curve to evaluate the discriminant ability of the model in distinguishing different types of temporal lobe epilepsy using the receiver operating characteristic curve;

[0010] (5) Draw the calibration curve of the model, and compare the consistency between the predicted probability and the actual observation by Bootstrap resampling.

[0011] (6) Draw the decision curve to analyze the clinical application benefit of the model under different threshold probabilities.

[0012] The data in step (1) include demographic data, clinical manifestations, vEEG, MRI, multi-modal cognitive and emotional scales, blood and cerebrospinal fluid tests, treatment regimens, and follow-up prognosis.

[0013] The binary Logistic regression analysis process in step (2) is as follows: for each candidate clinical feature x i , a univariate model is established:

[0014] Significance determination: if p < a (usually a = 0.05), x i is included in the candidate set for multi-factor analysis;

[0015] Suppose that after single-factor screening, k candidate variables x1, …, x k are selected, and a multivariate model is established:

[0016] The significance (p value) of each β j is used to determine whether the variable is an independent risk factor or an independent protective factor:

[0017] If β j > 0 and p < a, it is an independent risk factor (OR j = exp(β j )> 1);

[0018] If β j < 0 and p < a, it is an independent protective factor (OR j = exp(β j )< 1).

[0019] The establishment process of the nomogram model in step (3) is as follows: the linear predictor of the nomogram LP = β0+ β1x1+ β2x2+ … + β k x k , and the event probability is calculated: The nomogram converts the contribution of each variable into points, all points are added to obtain the total score (Total Points), and then the total score is mapped to the corresponding probability to obtain individual risk prediction.

[0020] The established nomogram model of step (3) takes the psychobehavioral disorders PBD, the seizure type FBTCS, and the normal blood neutrophil count NEUN as the prediction factors of the model, and the total score is 120 points; in the nomogram, the first behavior score scale, the score range is 0-100; the second behavior prediction probability, the range is 0-1; the third behavior psychobehavioral disorders PBD, 0 scale corresponds to 0 points, and 1 scale corresponds to 10 points; the fourth behavior seizure type FBTCS, 0 scale corresponds to 0 points, and 1 scale corresponds to 1 point; the fifth behavior normal blood neutrophil count NEUN, 0 scale corresponds to 0 points, and 1 scale corresponds to 1 point; the sixth behavior total score, the score range is 0-280 points; the seventh behavior linear prediction value, the range is-2 to 3; and the eighth behavior risk probability value, the score range is 0.3-0.9. The beneficial effects of the present application are as follows: based on the large sample clinical data of antibody-mediated autoimmune encephalitis patients, the present application systematically integrates demographic information, clinical manifestations, vEEG, MRI imaging features, cognitive and emotional scales, laboratory examinations and follow-up results, adopts the single factor and multi-factor binary Logistic regression method, screens out the risk factors and protective factors with independent prediction value, and establishes a nomogram prediction model. The established nomogram model shows high prediction efficiency and clinical practical value in the ROC curve, the correction curve and the decision curve analysis, and can realize individualized, quantitative and visualized risk assessment. In summary, the method provided by the present application can provide a scientific basis for early prognosis judgment and chronic epilepsy risk prediction of antibody-mediated autoimmune encephalitis patients, and provide strong support for clinical precise diagnosis and treatment and intervention, and has high popularization and application value. BRIEF DESCRIPTION OF DRAWINGS

[0021] Figure 1 It is the inclusion flowchart of different antibody-mediated AE patients.

[0022] Figure 2 It is the ROC curve analysis method for predicting the prognosis of antibody-mediated AE.

[0023] Figure 3 It is the method of the nomogram model for predicting the prognosis of antibody-mediated AE.

[0024] Figure 4 It is the method of the correction curve of the nomogram model for predicting the prognosis of antibody-mediated AE.

[0025] Figure 5 It is the method of the decision curve of the nomogram model for predicting the prognosis of antibody-mediated AE. DETAILED DESCRIPTION

[0026] For the purpose of facilitating the understanding of the present application, a more comprehensive description of the present application will be provided below. However, the present application can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present application more thorough and comprehensive.

[0027] Example 1

[0028] Research object: Collect the patients with antibody-mediated AE diagnosed in the Department of Neurology, the First Affiliated Hospital of Dalian Medical University from January 2016 to April 2024.

[0029] 1. Inclusion and exclusion criteria of the research object

[0030] Inclusion criteria: In accordance with the consensus criteria established by Graus et al. in 2016: (1) acute or subacute onset, with one or more symptoms such as limbic system symptoms, encephalitis syndrome, basal ganglia and / or midbrain / hypothalamic involvement, mental disorders, etc.; (2) one or more of the following auxiliary examination findings: ① Cerebrospinal fluid abnormalities: cerebrospinal fluid leukocytosis (> 5 × 10 6 / L), or cerebrospinal fluid cytology shows lymphocytic inflammation, or specific oligoclonal band is positive. ② Neuroimaging or electrophysiological abnormalities: cranial magnetic resonance imaging (MRI) limbic system T2 fluid-attenuated inversion recovery sequence (FLAIR) abnormal signal, unilateral or bilateral, or T2 FLAIR abnormal signal in other areas (except non-specific white matter changes and stroke); or PET limbic system high metabolic change, or multiple cortical and / or basal ganglia hypermetabolism. Electroencephalogram shows focal epilepsy or epileptiform discharge, or diffuse or multifocal slow wave rhythm; ③ Tumor related to AE; (3) Anti-neuron / glial cell antibody positive in serum and / or cerebrospinal fluid of the patient; (4) Exclude other causes. The enrolled patients have complete clinical data including clinical manifestations, imaging, video electroencephalogram (vEEG), lumbar puncture cerebrospinal fluid and blood tests, treatment and prognosis, etc., and the patients / families can provide written informed consent.

[0031] Exclusion criteria: (1) Not in line with the clinical diagnostic criteria of AE; (2) Patients lost to follow-up or died; (3) Incomplete clinical data; (4) Have other causes of seizures, including structural, genetic, infectious or metabolic factors; (5) Combined with other clear serious internal medicine diseases or serious mental and psychological diseases; combined with other clear central nervous system diseases such as cerebrovascular disease, spinal cord lesion, intracranial infection, intracranial space-occupying lesion, etc.; have had craniocerebral trauma or surgery; (6) Patients and families are unable or unwilling to provide written informed consent.

[0032] 2. Grouping of the research object:

[0033] Prognosis good and prognosis poor group: According to the modified Rankin Scale (mRS) (see Table 1), the prognosis was evaluated, and the mRS < 2 points after 1 year of follow-up after discharge were classified as the prognosis good group; the mRS ≥ 2 points were classified as the prognosis poor group.

[0034] Table 1 Modified Rankin (mRS) score scale

[0035]

[0036] The demographic data (including age, gender, disease duration), clinical manifestations, MRI, long-term vEEG, cognitive and emotional scale, blood and cerebrospinal fluid tests, treatment and prognosis of all patients were collected according to the electronic medical records (EMRs) of the patients. The cerebrospinal fluid and serum samples were sent to Shenyang Jinwei Testing Company or Nanjing Xiansheng Medical Testing Laboratory for AE-related antibody detection.

[0037] 1. General data collection

[0038] The following variables were extracted from the EMRs of the patients: (1) demographic information: age and gender; (2) disease duration, ICU admission; (3) clinical manifestations: main initial symptoms, seizure types, and AE-related clinical features (mental and behavioral disorders, emotional disorders, cognitive disorders, and consciousness disorders); (4) complications and comorbidities: prodromal infection, pneumonia, embolism events (pulmonary embolism and lower extremity arterial embolism), tumors, and other autoimmune diseases; (5) cerebrospinal fluid examination (pressure, white blood cell count, and protein quantification); laboratory tests: blood routine (platelets, white blood cells, neutrophils, lymphocytes), neutrophil-to-lymphocyte ratio (NLR), serum albumin, fasting blood glucose, blood sodium, inflammation-related markers [erythrocyte sedimentation rate (ESR) and C-reactive protein (CRP)]; (6) blood and cerebrospinal fluid autoimmune encephalitis, demyelination series, and paraneoplastic-related antibody detection; (7) treatment and prognosis: immunotherapy regimen, ASMs, etc.

[0039] 2. Multimodal MRI data collection

[0040] In this experiment, Philips Ingenia CX 3.0T MRI scanner, head 32-channel phased array coil, and American GE Signa 1.5T echospeed MRI scanner, orthogonal head coil were used for scanning in the magnetic resonance room of the First Affiliated Hospital of Dalian Medical University. The scanning sequences included: axial 3D-T1 weighted image (3D-T1WI), T2 weighted image (T2WI), T2 FLAIR, etc.

[0041] 3. Long-term vEEG data collection and electroencephalogram severity grading

[0042] The study used 32-channel Japanese light computer electrography EEG-1200C, using international 10-20 system scalp EEG electrode placement, select bilateral ear electrode as reference electrode, lead position is FP1, FP2, F3, F4, C3, C4, P3, P4, O1, O2, F7, F8, T3, T4, T5, T6, Fz, Cz, FPz. The resistance of each lead is <5kΩ, the sampling frequency is 200Hz, the filter range is 1-70Hz, the sensitivity is 100μV / cm, the time reference is 30mm / s, and the monitoring time is 2 hours or 24 hours. Long-term vEEG is interpreted by two experienced electroencephalogram doctors, and the electroencephalogram severity classification (see Table 2) and interictal epileptiform discharge (IEDs) index (semi-quantitative) (see Table 3) are determined. The electroencephalogram severity classification standard is to divide the acute phase EEG into normal, borderline, mild, moderate, severe and extremely severe electroencephalogram according to the background activity and paroxysmal abnormalities; the interictal epileptiform discharge index (semi-quantitative) is a semi-quantitative visual inspection according to the interictal epileptiform discharge, including occasional, small amount, slightly more, more, large amount and continuous.

[0043] Table 2 Electroencephalogram severity classification standard

[0044]

[0045] Abbreviation: IEDs: interictal epileptiform discharge

[0046] Table 3 Electroencephalogram interictal discharge index (semi-quantitative) calculation

[0047]

[0048] Abbreviation: EEG: electroencephalogram; vEEG: video electroencephalogram, ICU: intensive care unit.

[0049] 4. Modified Rankin (mRS) score scale

[0050] The modified Rankin (mRS) score scale was used to evaluate the severity of the patient's acute phase, and telephone follow-up was performed every 6 months after discharge. The prognosis was evaluated again by scoring with the table, and the follow-up results after 1 year of discharge were used as the standard. mRS <2 points is a good prognosis group; mRS ≥2 points is a poor prognosis group. At the same time, the efficacy evaluation group was divided, and the mRS score during follow-up was reduced by ≥2 points compared with the mRS score in the acute phase, which was the treatment effective group; the mRS score was reduced by 1 point, which was the treatment effective group; the mRS score was not reduced or increased, which was the treatment ineffective group.

[0051] 5. Statistical analysis

[0052] R (3.4.3) language and Graphpad Prism 9.0 software were used for statistical analysis and chart drawing. Measurement data were expressed as mean ± standard deviation or median, and count data were presented as frequency and constituent ratio. For the comparison of measurement data between groups that met the normal distribution, t-test was used. For the comparison of measurement data between groups that did not meet the normal distribution and the comparison of ordinal multiple classification variables between groups, Kruskal-Wallis H rank sum test was used. For the comparison of paired measurement data that did not meet the normal distribution, two related sample Wilcoxon signed rank test was used. Spearman rank correlation method was used for correlation analysis. The clinical characteristics were analyzed by single factor and multiple factor logistics regression analysis to screen out independent predictors of poor prognosis and conversion to chronic epilepsy. Receiver operating characteristic (ROC) curve was drawn, and area under the curve (AUC) was calculated to evaluate the predictive performance of each index. The weight coefficient of the independent predictor selected by the binary logistics regression model was used to establish a nomogram model, and the model calibration curve was drawn to intuitively evaluate the prediction accuracy of the model. Decision curve analysis (DCA) was used to evaluate the clinical application value of the model. P<0.05 was considered statistically significant.

[0053] Results:

[0054] 1. General information: From January 2016 to April 2024, 178 potential AE patients were screened by searching EMRs, as shown in Figure 1 According to the exclusion criteria, 94 patients were excluded, including 57 patients with “possible AE”, 4 patients who died or were lost to follow-up, 25 patients with incomplete clinical data, and 8 patients with other causes of seizures. Finally, 84 patients with “definite antibody-mediated AE” were included. Among them, 35 cases (41.6%) were anti-LGI1 encephalitis, 16 cases (19.0%) were anti-NMDAR encephalitis, 12 cases (14.3%) were anti-GABABR encephalitis, 3 cases (3.6%) were anti-Caspr2 encephalitis, 10 cases (11.9%) were anti-GAD65 encephalitis, 5 cases (6.0%) were MOG antibody cortical encephalitis (FLAMES), and 3 cases (3.6%) were GFAP-A encephalitis.

[0055] According to the follow-up of whether there was a seizure after discharge for 1 year, 23 cases were divided into the chronic epilepsy group and 49 cases were divided into the non-chronic epilepsy group. At the same time, 84 AE patients were divided into the good prognosis group (43 cases) and the poor prognosis group (41 cases) according to the follow-up score <2 or ≥2 after discharge for 1 year.

[0056] 2. Risk prediction of poor prognosis

[0057] 2.1 Single factor analysis

[0058] In order to identify the prognosis of antibody-mediated AE early, find the potential risk factors of poor prognosis, 84 cases of antibody-mediated AE patients were divided into 2 groups: good prognosis group (43 cases) and poor prognosis group (41 cases), the clinical data of 2 groups were compared between groups, the results of single factor analysis showed that: the main initial symptoms were disturbance of consciousness (2.33% vs 19.51%, p<0.028), disturbance of consciousness during the course of the disease (11.63% vs 43.90%, p<0.01), mental and behavioral abnormalities (32.56% vs 60.98%, p<0.01), GTCS type of seizure (51.16% vs 75.61%, p<0.02), FBTCS (20.93% vs 53.66%, p<0.01), SE during the course of the disease (13.95% vs 36.59%, p<0.017), EEG background showed diffuse slow wave (11.63% vs 31.71%, p<0.025), EEG showed continuous abnormal discharge (2.33% vs 19.51%, p<0.028), EEG severity classification was severe or extremely severe abnormality (27.91% vs 51.22%, p<0.029), elevated cerebrospinal fluid pressure (p<0.029), cerebrospinal fluid protein content (p<0.024), elevated ESR (16.28% vs 39.02%, p<0.019), combined with other antibodies (serum) (20.93% vs 43.90%, p<0.024), acute phase mRS score≧3 (83.72% vs 100%, p<0.021), suggesting that the above factors are related to poor prognosis; while EEG severity classification was normal or borderline (27.91% vs 9.76%, p<0.034), normal blood neutrophil count (81.40% vs 58.54%, p<0.022), normal blood lymphocyte count (83.72% vs 63.41%, p<0.034), high NLR (20.93% vs 43.90%, p<0.024), normal NLR (79.07% vs 53.66%, p<0.014), may indicate good prognosis (13.04% vs 4.08%, p<0.018); the differences of the above factors in 2 groups of antibody-mediated AE patients were statistically significant (p<0.05), which may become independent factors for predicting the prognosis of antibody-mediated AE. See Tables 4-7.

[0059] Table 4 Comparison of clinical characteristics between antibody-mediated AE with good prognosis and poor prognosis

[0060]

[0061]

[0062] Table 5 Comparison of clinical features between antibody-mediated AE good and poor prognosis groups (continued)

[0063]

[0064]

[0065] Table 6 Comparison of clinical features between antibody-mediated AE good and poor prognosis groups (continued)

[0066]

[0067]

[0068] Table 7 Comparison of clinical features between antibody-mediated AE good and poor prognosis groups (continued)

[0069]

[0070]

[0071] FBTCS: focal onset secondary to generalized tonic-clonic seizure; SE: status epilepticus; MOCA: Montreal Cognitive Assessment; MMSE: Mini Mental State Examination; HAMA: Hamilton Anxiety Scale; HAMD: Hamilton Depression Scale; ESR: erythrocyte sedimentation rate; NLR: neutrophil lymphocyte ratio; CRp: C-reactive protein; IVIG: intravenous immunoglobulin; ASMs: antiseizure medications; mRS: modified Rankin Scale; IVIG: intravenous immunoglobulin, *: p<0.05; **: p<0.01.

[0072] 2.2 Binary logistics regression analysis: First, for each candidate clinical feature x i which can be binary, ordinal or continuous, a univariate model is established:

[0073] Significance determination: if p< a (commonly a = 0.05) then x i is included in the candidate set for multivariate analysis;

[0074] Suppose after univariate screening, k candidate variables x1, …, x k are selected, a multivariate model is established:

[0075] The significance (p-value) of each β j is used to determine whether the variable is an independent risk factor or an independent protective factor:

[0076] If β j > 0 and p < a, it is an independent risk factor (OR j = exp(β j > 1);

[0077] If β j < 0 and p < a, it is an independent protective factor (OR j = exp(β j < 1).

[0078] Therefore, in this example, all factors with p < 0.05 in single factor analysis and factors found in previous studies that are significantly associated with the prognosis of antibody-mediated AE were selected, including: the presence of consciousness disorder during the course, PBD, GTCS, FBTCS, SE during the course, MoCA scale score, EEG background showing diffuse slow wave, EEG showing continuous abnormal discharge, EEG severity classification as normal or borderline, EEG severity classification as moderate or extremely severe abnormality, cerebrospinal fluid pressure value, cerebrospinal fluid protein content, normal blood neutrophil count, normal blood lymphocyte count, high blood NLR, normal blood NLR, elevated ESR, combined with other antibodies, linear analysis, excluding factors with variance inflation factor (VIF) > 10 in linear analysis, and binary logistics regression analysis of the selected factors, see Tables 10-11. The results show that: PBD during the course (OR = 5.98, 95% CI: 1.21-29.52, p < 0.028) and FBTCS (OR = 7.21, 95% CI: 1.13-46.21, p < 0.037) are independent risk factors for poor prognosis of antibody-mediated AE, and normal blood neutrophil count (OR = 0.09, 95% CI: 0.01-0.91, p < 0.042) is an independent protective factor for good prognosis, see Table 8.

[0079] Table 8 Binary logistics regression analysis for predicting the prognosis of antibody-mediated AE

[0080]

[0081]

[0082] Abbreviations: GTCS: general tonic-clonic seizure; FBTCS: focal seizure secondary to general tonic-clonic seizure; SE: status epilepticus; MOCA: Montreal Cognitive Assessment Scale; EEG: electroencephalogram; NLR: neutrophil-to-lymphocyte ratio; *: 0.01 ≤ p < 0.05; **: 0.001 ≤ p < 0.01; ***: p < 0.001.

[0083] 2.3 ROC curve analysis

[0084] The ROC curve analysis of the three independent factors affecting the prognosis of antibody-mediated AE (PBD, FBTCS, and NEUN) showed that the AUC value of PBD was 0.642 (95% CI: 0.523-0.761), the AUC value of FBTCS was 0.664 (95% CI: 0.546-0.781), and the AUC value of NEUN was 0.614 (95% CI: 0.264-0.507). Delong test showed that the performance of the prediction model was better than that of each single factor (all p<0.05), as shown in Table 9. Figure 2 .

[0085] Table 9 Comparison of the performance of independent prediction factors and prediction models for predicting the prognosis of antibody-mediated AE

[0086]

[0087] Abbreviations: FBTCS: focal seizure secondary to generalized tonic-clonic seizure; 95% CI: 95% confidence interval; a: represents the delong test result; Na: no statistics.

[0088] Figure 2 ROC curve analysis for predicting the prognosis of antibody-mediated AE: ROC curves were drawn for the three independent factors affecting the prognosis of antibody-mediated AE (PBD, FBTCS, and NEUN). The AUC value of PBD was 0.642 (95% CI: 0.523-0.761), the AUC value of FBTCS was 0.664 (95% CI: 0.546-0.781), and the AUC value of NEUN was 0.614 (95% CI: 0.264-0.507). The AUC value of the prediction probability was 0.915 (95% CI: 0.834-0.965), indicating that the performance of the prediction model was good. Blue curve: PBD; green curve: FBTCS; red curve: NEUN; purple dotted line: prediction probability.

[0089] 2.4 Construction of nomogram model for predicting the prognosis of antibody-mediated AE

[0090] The R language is used to screen the independent predictive factor weight coefficient according to the binary logistics regression model, and a nomogram model for predicting the prognosis of antibody-mediated AE is established. The establishment process is: the linear prediction factor LP of the nomogram is β0+β1x1+β2x2+…+β k x k , the event occurrence probability is calculated: The nomogram converts the contribution of each variable into points, all points are accumulated to obtain the total score, and then the total score is mapped to the corresponding probability to obtain individualized risk prediction. Therefore, in this example, the mental behavior disorder PBD, the seizure type FBTCS and the normal blood neutrophil count NEUN are used as the prediction factors for constructing the model, and the coefficients obtained according to the regression model are β0=-0.63, β1=1.75, β2=1.81 and β3=-1.32, respectively. Therefore, the linear prediction factor expression of the model is LP=-0.63+1.75*PBD+1.81*FBTCS+-1.32*NEUN. The regression coefficients of each prediction variable in the model are linearly mapped to the corresponding points (Points). PBD and FBTCS have larger coefficients and make more significant contributions to the total score; and the negative coefficient of NEUN is reduced in the nomogram. The total score of the individual can be obtained by accumulating the points of all variables, and then the individualized risk value of the patient for adverse prognosis is obtained according to the mapping relationship between the total score and the prediction probability, as shown in Figure 3 The model calibration curve is drawn, and it can be seen that the curve is basically close to the ideal curve, indicating that the prediction degree of the model is good, as shown in Figure 4 The decision curve shows that when the risk threshold is <0.8, the clinical net benefit provided by the nomogram model is good, as shown in Figure 5 .

[0091] The above-described embodiments only express several embodiments of the present application, and the description is more specific and detailed, but it cannot be understood as limiting the scope of the patent. It should be noted that for ordinary skilled persons in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are within the scope of protection of the present application. Therefore, the protection scope of the patent of the present application should be subject to the appended claims.

Claims

1. A method for establishing a nomogram model for prognostic assessment of antibody-mediated autoimmune encephalitis, characterized in that, Follow these steps: (1) Collect data from patients with antibody-mediated autoimmune encephalitis; (2) Univariate and multivariate binary logistic regression analyses were performed on the included clinical characteristics to screen for independent risk factors and protective factors; (3) Establish a nomogram model based on the regression coefficients of the binary logistic regression model to quantify the contribution of each predictor to the prognosis; (4) Plot ROC curves and use receiver operating characteristic curves to evaluate the model’s ability to distinguish different types of temporal lobe epilepsy; (5) Plot the model calibration curve, use Bootstrap repeated sampling, and compare the consistency between the model's predicted probability and the actual observation results; (6) Plot decision curves and use them to analyze the clinical application benefits of the model under different threshold probabilities.

2. The method for establishing a nomogram model for prognostic assessment of antibody-mediated autoimmune encephalitis according to claim 1, characterized in that, The data in step (1) includes demographic information, clinical manifestations, vEEG, MRI, multimodal cognitive and emotional scales, blood and cerebrospinal fluid tests, treatment plans and follow-up prognosis.

3. The method for establishing a nomogram model for prognostic assessment of antibody-mediated autoimmune encephalitis according to claim 1, characterized in that, The binary logistic regression analysis process described in step (2) is as follows: for each candidate clinical feature x i The model can be binary, ordered, or continuous; a univariate model can be established. Significance test: If p < α (α is usually taken as 0.05), then x i Candidate set to be included in multivariate analysis; Suppose that k candidate variables x1,…,x are selected after single-factor screening. k Establish a multivariate model: Each β j The significance (p-value) is used to determine whether the variable is an independent risk factor or a protective factor: If β j If the value is greater than 0 and p < α, then it is an independent risk factor (OR). j =exp(β) j )>1); If β j If p < 0 and p < α, then it is an independent protective factor (OR). j =exp(β) j <1).

4. The method for establishing a nomogram model for prognostic assessment of antibody-mediated autoimmune encephalitis according to claim 1, characterized in that, The process of establishing the nomogram model in step (3) is as follows: The linear predictor of the nomogram is LP = β0 + β1x1 + β2x2 + ... + β k x k Calculation of the probability of an event occurring: Nodal plots convert the contribution of each variable into points, and all points are summed to obtain a total score. The total score is then mapped to the corresponding probability to obtain individualized risk prediction.

5. The method for establishing a nomogram model for prognostic assessment of antibody-mediated autoimmune encephalitis according to claim 1, characterized in that, The nomogram model established in step (3) uses mental and behavioral disorders, epileptic seizure type FBTCS, and normal blood neutrophil count as predictive factors for model construction, with a total score of 120 points; in the nomogram, the first row is a score scale with a score range of 0 to 100; the second row is the predicted probability with a range of 0 to 1; the third row is the mental and behavioral disorder PBD, with a scale of 0 corresponding to 0 points and a scale of 1 corresponding to 10 points; The fourth line represents the seizure type FBTCS, with 0 on the scale corresponding to 0 points and 1 on the scale corresponding to 1 point. The fifth line shows a normal neutrophil count in the blood (NEUN). A score of 0 corresponds to 0 on the scale bar, and a score of 1 corresponds to 1 on the scale bar. The sixth row is the total score, ranging from 0 to 280 points; the seventh row is the linear prediction value, ranging from -2 to 3; and the eighth row is the risk probability value, ranging from 0.3 to 0.9.

Images