Clinical data processing method and device, computer device, readable storage medium and program product

Abstract

This application relates to a method, apparatus, computer device, computer-readable storage medium, and computer program product for processing clinical data. The method includes: acquiring clinical data corresponding to a target object identifier; the clinical data includes memory and cognitive assessment results and structural data of a target brain region; the target brain region is a brain region associated with pathological features of Alzheimer's disease; inputting the memory and cognitive assessment results and the structural data of the target brain region into a pre-trained glucose uptake rate determination model, and outputting the glucose uptake rate of the posterior cingulate cortex corresponding to the target object identifier; classifying the target object identifier into a first type or a second type based on the glucose uptake rate of the posterior cingulate cortex; wherein the glucose metabolism level of the posterior cingulate cortex corresponding to an object identifier belonging to the first type is superior to the glucose metabolism level of the posterior cingulate cortex corresponding to an object identifier belonging to the second type. This method can improve the classification efficiency for objects with memory and cognitive impairment.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a method, apparatus, computer device, computer-readable storage medium, and computer program product for processing clinical data. Background Technology

[0002] Alzheimer's disease (AD), also known as senile dementia, is a degenerative disease of the central nervous system. It has an insidious onset and a chronic, progressive course, and is the most common type of dementia in the elderly. The main pathological features of AD include the formation of neurofibrillary tangles caused by amyloid β (Aβ) plaques and abnormal phosphorylation of Tau protein. These pathological processes work together to lead to synaptic dysfunction and neurodegeneration.

[0003] Currently, monoclonal antibodies, such as lencanezumab, have been proven to effectively slow cognitive decline during the progression of Alzheimer's disease (AD). However, individual variability in clinical response remains a significant challenge—even with Aβ clearance, some patients experience only minimal cognitive deterioration. One of the core goals of precision medicine is to predict an individual's potential for functional recovery after treatment intervention by identifying reliable biomarkers, thereby achieving individual stratification.

[0004] The medial temporal lobe (including the hippocampus) is one of the earliest brain regions affected and atrophied by neurofibrillary tangles. Its structural integrity can be assessed using the Medial Temporal Atrophy (MTA) scale. The MTA is a scale that uses MRI or CT images to visually score the degree of atrophy in structures such as the entorhinal cortex and hippocampus from 0 to 4. It can quantify the structural integrity of the medial temporal lobe and can serve as a key indicator reflecting hippocampal integrity.

[0005] The posterior cingulate cortex (PCC) is a key vulnerable area, and impaired glucose metabolism is considered an important biomarker. Fluorodeoxyglucose-positron emission tomography (FDG-PET) is the gold standard technique for quantifying glucose metabolism in local brain regions and identifying these prognostic low-metabolic patterns, but its application is limited due to its high cost, complex operation, and exposure to ionizing radiation, affecting the efficiency of individual stratification in AD treatment. Summary of the Invention

[0006] Therefore, it is necessary to provide a method, apparatus, computer device, computer-readable storage medium, and computer program product for processing clinical data that can improve the classification efficiency of objects with memory and cognitive impairment, in order to address the above-mentioned technical problems.

[0007] Firstly, this application provides a method for processing clinical data, including:

[0008] Acquire clinical data corresponding to the target object identifier; the clinical data includes memory and cognitive assessment results and structural data of the target brain region; the target brain region is a brain region associated with the pathological features of Alzheimer's disease; the target object identifier is the object identifier corresponding to the target object with memory and cognitive impairment;

[0009] The memory and cognitive assessment results and the structural data of the target brain region are input into a pre-trained glucose uptake rate determination model, which outputs the glucose uptake rate of the posterior cingulate cortex corresponding to the target object identifier. The pre-trained glucose uptake rate determination model is obtained by performing correlation analysis between the memory and cognitive assessment test results corresponding to the subject identifier and the structural test data of the target brain region and different brain regions corresponding to the subject identifier. The subject identifier is the object identifier corresponding to the subject with memory and cognitive impairment.

[0010] Based on the glucose uptake rate of the posterior cingulate cortex, the target object is classified into a first type or a second type; wherein, the glucose metabolism level of the posterior cingulate cortex corresponding to the object object belonging to the first type is better than the glucose metabolism level of the posterior cingulate cortex corresponding to the object object belonging to the second type.

[0011] In one embodiment, the clinical data further includes object attribute data corresponding to the target object identifier in at least one object attribute; the step of inputting the memory and cognitive assessment results and the structural data of the target brain region into a pre-trained glucose uptake rate determination model, and outputting the glucose uptake rate of the posterior cingulate cortex corresponding to the target object identifier, includes:

[0012] The memory and cognition assessment results, the structural data of the target brain region, and the object attribute data are input into the pre-trained glucose uptake rate determination model, and the glucose uptake rate of the posterior cingulate cortex corresponding to the target object identifier is output.

[0013] The glucose uptake rate is obtained by the pre-trained glucose uptake rate determination model, which processes the memory and cognition assessment results, the structural data of the target brain region, and the object attribute data according to the weight information corresponding to different clinical data; the weight information represents the relative importance of the corresponding clinical data to the glucose metabolism analysis of the posterior cingulate cortex.

[0014] In one embodiment, the target brain region includes the medial temporal lobe, and the acquisition of clinical data corresponding to the target object identifier includes:

[0015] Obtain structural evaluation data of the medial temporal lobe corresponding to the target object identifier; the structural evaluation data is used to characterize the structural integrity of the medial temporal lobe.

[0016] The structural assessment data of the medial temporal lobe corresponding to the target object identifier is used as the structural data of the target brain region.

[0017] In one embodiment, obtaining the structural evaluation data of the medial temporal lobe corresponding to the target object identifier includes:

[0018] Obtain the atrophy score of the medial temporal lobe corresponding to the target object identifier;

[0019] The atrophy score is used as the structural assessment data.

[0020] In one embodiment, the memory and cognition assessment result includes the MemTrax assessment result obtained by performing a memory and cognition assessment based on the MemTrax system; the MemTrax assessment result includes at least the memory and cognition accuracy rate.

[0021] In one embodiment, classifying the target object into a first type or a second type based on the glucose uptake rate of the posterior cingulate cortex includes:

[0022] Obtain a preset glucose metabolism threshold; the glucose metabolism threshold is a cutoff value determined based on the Youden index by performing ROC curve analysis on the pre-trained glucose uptake rate determination model.

[0023] If the glucose uptake rate in the posterior cingulate cortex meets the glucose metabolism threshold, the target object is identified and classified as the first type.

[0024] If the glucose uptake rate in the posterior cingulate cortex does not meet the glucose metabolism threshold, the target object is identified and classified as the second type.

[0025] Secondly, this application also provides a clinical data processing device, comprising:

[0026] The acquisition module is used to acquire clinical data corresponding to the target object identifier; the clinical data includes memory and cognitive assessment results and structural data of the target brain region; the target brain region is a brain region associated with the pathological features of Alzheimer's disease; the target object identifier is the object identifier corresponding to the target object with memory and cognitive impairment;

[0027] The prediction module is used to input the memory and cognitive assessment results and the structural data of the target brain region into a pre-trained glucose uptake rate determination model, and output the glucose uptake rate of the posterior cingulate cortex corresponding to the target object identifier; the pre-trained glucose uptake rate determination model is obtained by performing correlation analysis between the memory and cognitive assessment test results corresponding to the subject identifier and the structural test data of the target brain region and different brain regions corresponding to the subject identifier; the subject identifier is the object identifier corresponding to the subject with memory and cognitive impairment.

[0028] The classification module is used to classify the target object identifier into a first type or a second type based on the glucose uptake rate of the posterior cingulate cortex; wherein the glucose metabolism level of the posterior cingulate cortex corresponding to the object identifier belonging to the first type is better than the glucose metabolism level of the posterior cingulate cortex corresponding to the object identifier belonging to the second type.

[0029] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor, the memory storing a computer program that, when executed by the processor, implements the steps of the method described above.

[0030] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements the steps of the above-described method.

[0031] Fifthly, this application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, implements the steps of the above-described method.

[0032] The aforementioned clinical data processing methods, devices, computer equipment, computer-readable storage media, and computer program products acquire clinical data corresponding to target object identifiers. The clinical data includes memory and cognitive assessment results and structural data of the target brain region. The target brain region is a brain region associated with the pathological characteristics of Alzheimer's disease. The target object identifier is the object identifier corresponding to a target object exhibiting memory and cognitive impairment. The memory and cognitive assessment results and the structural data of the target brain region are input into a pre-trained glucose uptake rate determination model, which outputs the glucose uptake rate of the posterior cingulate cortex corresponding to the target object identifier. The pre-trained glucose uptake rate determination model is obtained by performing correlation analysis between the memory and cognitive assessment test results corresponding to the subject identifier and the structural test data of the target brain region, and different brain regions corresponding to the subject identifier. The subject identifier is the object identifier corresponding to a subject exhibiting memory and cognitive impairment. Based on the glucose uptake rate of the posterior cingulate cortex, the target object identifier is classified into a first type or a second type. The glucose metabolism level of the posterior cingulate cortex corresponding to a subject identifier belonging to the first type is superior to the glucose metabolism level of the posterior cingulate cortex corresponding to a subject identifier belonging to the second type.

[0033] Thus, by integrating pre-trained glucose uptake determination models that combine memory and cognitive assessment test results with structural test data of the target brain region, glucose uptake in the posterior cingulate cortex of the target subject can be accurately predicted based on clinical data containing both memory and cognitive assessment results and structural data of the target brain region. This transforms readily available clinical data into reliable surrogate indicators of key but difficult-to-obtain neuroimaging biomarkers, going beyond simple statistical correlation descriptions. It provides an operational model for a core pathophysiological concept: integrating memory and cognitive assessment results related to the posterior cingulate cortex, the network hub of the default mode network, with its primary input nodes. Structural data of brain regions associated with the pathological features of Alzheimer's disease are used to assess the functional status of this network hub. This allows for a low-cost, non-invasive alternative to FDG-PET to predict the metabolic state of the posterior cingulate cortex of a target individual. By utilizing the metabolic state of the posterior cingulate cortex, the target individual can be efficiently classified into type I or type II based on their glucose metabolism level. Since the glucose metabolism level of the posterior cingulate cortex is closely related to the therapeutic effect of drugs used to treat memory and cognitive impairment, this method can efficiently stratify patients before drug treatment, improving classification efficiency and enabling precision medicine. Attached Figure Description

[0034] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0035] Figure 1 This is a flowchart illustrating a clinical data processing method in one embodiment;

[0036] Figure 2 This is a flowchart illustrating a clinical data processing method in another embodiment;

[0037] Figure 3(A) is a correlation diagram of posterior cingulate cortex metabolism, cognitive function and structural integrity in one embodiment;

[0038] Figure 3(B) is a correlation diagram of posterior cingulate cortex metabolism and cognitive function in another embodiment;

[0039] Figure 3(C) is a correlation diagram of the metabolic values ​​of the posterior cingulate cortex and the metabolic images of the posterior cingulate cortex in another embodiment;

[0040] Figure 3(D) shows a correlation between posterior cingulate cortex metabolism and cognitive level in another embodiment;

[0041] Figure 4(A) is a schematic diagram of the parameter correlation of a model training set in one embodiment;

[0042] Figure 4(B) is a schematic diagram of a model verification using a test set in one embodiment;

[0043] Figure 4(C) is a schematic diagram of the relative importance of predictors in a model in one embodiment;

[0044] Figure 4(D) is a schematic diagram of ROC curve analysis for one model in one embodiment;

[0045] Figure 4(E) is a schematic diagram of a method for classifying metabolic values ​​in one embodiment;

[0046] Figure 5(A) is a schematic diagram of the correlation between a baseline predicted score and the total ADAS-Cog score in one embodiment;

[0047] Figure 5(B) is a schematic diagram of the correlation between a baseline prediction score and a word recall score in one embodiment;

[0048] Figure 5(C) is a schematic diagram of the correlation between a baseline prediction score and a word re-identification score in one embodiment;

[0049] Figure 6 This is a schematic diagram of a different cognitive change trajectory in one embodiment;

[0050] Figure 7 This is a structural block diagram of a clinical data processing device in one embodiment;

[0051] Figure 8 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0052] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0053] It should be noted that the terms "first," "second," etc., used in this application can be used to describe various elements, but these elements are not limited by these terms. These terms are only used to distinguish the first element from the second element. The terms "comprising" and "having," and any variations thereof, used in this application, are intended to cover non-exclusive inclusion. The term "multiple" used in this application refers to two or more. The term "and / or" used in this application refers to one of the embodiments, or any combination of multiple embodiments.

[0054] In one embodiment, such as Figure 1 As shown, a method for processing clinical data is provided. This embodiment illustrates the application of this method to a computer device. It is understood that the computer device may be a terminal, a server, or a system including both a terminal and a server. In this embodiment, the method includes the following steps:

[0055] Step S110: Obtain the clinical data corresponding to the target object identifier.

[0056] Among them, the target object identifier is the object identifier corresponding to the target object with memory and cognitive impairment. The target object with memory and cognitive impairment includes, but is not limited to, objects with mild cognitive impairment, mild AD, etc.

[0057] The target group can be individuals undergoing efficacy response stratification before receiving drug treatment. The drugs can include monoclonal antibodies, such as lencanezumab.

[0058] In some embodiments, individuals who are positive for amyloid-β, confirmed by PET-CT (Positron Emission Tomography / Computed Tomography), have a Clinical Dementia Rating (CDR) score of 0.5 or 1, and have been excluded from cognitive impairment caused by vascular cognitive impairment, mental disorders, and other diseases, can be identified as having memory and cognitive impairment. Amyloid-β is a neuronal structural protein associated with the pathogenesis of Alzheimer's disease (AD).

[0059] The clinical data includes memory and cognitive assessment results as well as structural data of the target brain regions.

[0060] Among them, the memory and cognitive assessment results are the result data obtained by assessing the memory and cognitive status of the target object, which are used to characterize the target object's memory and cognitive status.

[0061] In one embodiment, the target subject may undergo a memory cognitive assessment using a memory cognitive assessment system. A computer device can acquire the memory cognitive assessment results obtained by the target subject through the memory cognitive assessment system.

[0062] In some embodiments, the memory cognitive assessment results include MemTrax assessment results obtained based on the MemTrax system; the MemTrax assessment results include at least the memory cognitive accuracy rate.

[0063] The MemTrax system is a memory and cognitive assessment system. MemTrax provides target subjects with images that have a certain repetition rate for judgment. Based on the target subject's accuracy and reaction time in judging the repeated images, its cognitive function is predicted, yielding a MemTrax assessment result. The MemTrax assessment result may include, but is not limited to, MemTrax accuracy, cognitive reaction time, and the judgment result indicating whether cognitive function is at a normal level. It can be understood that memory and cognitive accuracy includes MemTrax accuracy (memory and cognitive accuracy). By limiting the memory and cognitive assessment results to include MemTrax assessment results, the predictive accuracy of glucose uptake rate in the posterior cingulate cortex of the target subject can be further improved.

[0064] The target brain regions are those associated with the pathological features of Alzheimer's disease (AD). The main pathological features of AD include the formation of neurofibrillary tangles caused by amyloid β plaques and abnormal phosphorylation of Tau protein. These pathological processes work together to lead to synaptic dysfunction and neurodegeneration.

[0065] In practice, computer devices can obtain relevant clinical data based on the target object identifier corresponding to the target object, including memory and cognitive assessment results and structural data of the target brain region.

[0066] Step S120: Input the memory and cognitive assessment results and the structural data of the target brain region into the pre-trained glucose uptake rate determination model, and output the glucose uptake rate of the posterior cingulate cortex corresponding to the target object identifier.

[0067] The posterior cingulate cortex (PCC) is a key hub in the pathological process of AD, and its glucose metabolism impairment is closely related to cognitive decline.

[0068] In the progression of Alzheimer's disease (AD), the posterior cingulate cortex (PCC) is a key vulnerable area, and impaired glucose metabolism is considered an important biomarker. As the core hub of the Default Mode Network (DMN), the PCC exhibits significant metabolic decline in the early stages of AD. This region is the first to show multiple pathological changes, including Aβ pathological deposition, structural atrophy, and metabolic decline, making it a key integrator of the early pathophysiological mechanisms of AD. Simultaneously, low PCC metabolism is closely associated with cognitive impairment, particularly in episodic memory. Functional neuroimaging studies have consistently confirmed that the PCC is an important brain network node supporting episodic memory processes. It has significant value in predicting disease progression, and its predictive efficacy has been validated in multiple different populations even in the preclinical stage. Besides its close association with memory function, the metabolic state of the PCC can also serve as a surrogate indicator of "neural reserve"—high metabolism implies a higher synaptic density retention rate, resulting in greater benefit from amyloid clearance; while low metabolism may suggest lower synaptic retention and a worse prognosis. In summary, PCC, as a key and multi-faceted biomarker in AD, possesses functional integrity, exhibiting early appearance, being closely associated with core memory deficits, and effectively predicting disease progression.

[0069] Early-stage Alzheimer's disease (AD) is not characterized by isolated PCC metabolic impairment, but rather by associated decline in hippocampal memory function. This association may stem from hippocampal atrophy or impaired integrity of its neural circuits. The medial temporal lobe (including the hippocampus) is one of the earliest brain regions to exhibit neurofibrillary tangles and atrophy in AD, and its early vulnerability makes it a key structural biomarker. Therefore, integrating the Medial Temporal Atrophy (MTA), a method for grading the degree of atrophy in the medial temporal lobe using MRI or CT imaging, can significantly improve the predictive ability of PCC glucose metabolism levels. In conclusion, the close structural and functional coupling between the PCC and the hippocampus constitutes the core of the early AD pathophysiological mechanism, suggesting the importance of combining multimodal biomarkers for accurate assessment of disease progression.

[0070] Accordingly, the target brain region includes the medial temporal lobe, and the structural data of the target brain region can refer to data obtained by assessing the structural integrity of the medial temporal lobe based on the MTA scoring method. In some embodiments, structural assessment data of the medial temporal lobe corresponding to the target object identifier can be obtained; the structural assessment data is used to characterize the structural integrity of the medial temporal lobe; the structural assessment data of the medial temporal lobe corresponding to the target object identifier is used as the structural data of the target brain region. In some embodiments, the atrophy score of the medial temporal lobe corresponding to the target object identifier can be obtained; the atrophy score is used as the structural assessment data. The atrophy score is the MTA score. The MTA scale (used to measure choroidal thickness) is a 5-point scale from 0 to 4, and its measurement is based on visual analysis of the choroidal fissure width, temporal horn, and hippocampal structural height. All MTA scale measurements are performed visually by a specialist physician.

[0071] The pre-trained glucose uptake rate determination model was obtained by correlation analysis between the memory and cognitive assessment test results corresponding to the subject identifiers and the structural test data of the target brain regions and the different brain regions corresponding to the subject identifiers. The subject identifiers were the object identifiers corresponding to subjects with memory and cognitive impairments (such as mild cognitive impairment, mild AD). The pre-trained glucose uptake rate model was a linear regression model.

[0072] Among the correlation analysis, the posterior cingulate cortex showed the strongest correlation with memory and cognitive assessment results and MTA scores across all brain regions.

[0073] In some embodiments, a pre-trained glucose uptake determination model is created by performing cross-sectional analysis on clinical trial data of subjects to establish the correlation between memory and cognitive assessment results (MemTrax assessment results), MTA scores (structural assessment data of the medial temporal lobe), and PCC glucose metabolism.

[0074] In specific implementation, the memory and cognitive assessment results corresponding to the target object identifier and the structural data of the target brain region are input into a pre-trained glucose uptake rate determination model, which outputs the glucose uptake rate of the posterior cingulate cortex corresponding to the target object identifier. In some embodiments, the MemTrax accuracy (MTx-%C) corresponding to the target object identifier and the structural assessment data (MTA score) of the medial temporal lobe can be input into the pre-trained glucose uptake rate determination model, which outputs the glucose uptake rate of the posterior cingulate cortex corresponding to the target object identifier.

[0075] Step S130: Based on the glucose uptake rate of the posterior cingulate cortex, the target object is identified and classified into a first type or a second type.

[0076] Among them, the glucose metabolism level of the posterior cingulate cortex corresponding to the first type of object identifier is better than the glucose metabolism level of the posterior cingulate cortex corresponding to the second type of object identifier.

[0077] In practical applications, the first type can be named the high metabolism prediction group, and the second type can be named the low metabolism prediction group.

[0078] In practice, target identifiers can be classified into a first type or a second type based on the glucose uptake rate in the posterior cingulate cortex corresponding to the target identifier. The glucose metabolism level in the posterior cingulate cortex corresponding to target identifiers of the first type is superior to that corresponding to target identifiers of the second type.

[0079] While lencanetumab can reduce amyloid burden, the conversion of this pathological change into cognitive benefit depends on the brain's residual functional capacity. Individuals in the high metabolic prediction group retain greater neural capacity—their PCC and its connectivity networks maintain the synaptic and functional plasticity required for adaptation. In contrast, the low metabolic prediction group may represent a state of failure or exhaustion of these compensatory mechanisms. That is, patients with a higher PCC SUVR (high metabolic group) possess the physiological basis needed to utilize the cleared pathological "space" for network stabilization or recovery, resulting in a superior trajectory of cognitive change. Conversely, exhausted patients (low metabolic group) lack this adaptive potential, limiting the clinical efficacy of amyloid clearance. Therefore, the high metabolic prediction group may achieve better therapeutic outcomes with monoclonal antibody therapy, represented by lencanetumab.

[0080] In the above-mentioned clinical data processing method, clinical data corresponding to the target object identifier is obtained. The clinical data includes memory and cognitive assessment results and structural data of the target brain region. The target brain region is a brain region associated with the pathological characteristics of Alzheimer's disease. The target object identifier is the object identifier corresponding to the target object with memory and cognitive impairment. The memory and cognitive assessment results and the structural data of the target brain region are input into a pre-trained glucose uptake rate determination model, which outputs the glucose uptake rate of the posterior cingulate cortex corresponding to the target object identifier. The pre-trained glucose uptake rate determination model is obtained by performing correlation analysis between the memory and cognitive assessment test results corresponding to the subject identifier and the structural test data of the target brain region and different brain regions corresponding to the subject identifier. The subject identifier is the object identifier corresponding to the subject with memory and cognitive impairment. According to the glucose uptake rate of the posterior cingulate cortex, the target object identifier is classified into a first type or a second type. Among them, the glucose metabolism level of the posterior cingulate cortex corresponding to the first type of object identifier is better than the glucose metabolism level of the posterior cingulate cortex corresponding to the second type of object identifier.

[0081] Thus, by integrating pre-trained glucose uptake determination models that combine memory and cognitive assessment test results with structural test data of the target brain region, glucose uptake in the posterior cingulate cortex of the target subject can be accurately predicted based on clinical data containing both memory and cognitive assessment results and structural data of the target brain region. This transforms readily available clinical data into reliable surrogate indicators of key but difficult-to-obtain neuroimaging biomarkers, going beyond simple statistical correlation descriptions. It provides an operational model for a core pathophysiological concept: integrating memory and cognitive assessment results related to the posterior cingulate cortex, the network hub of the default mode network, with its primary input nodes. Structural data of brain regions associated with the pathological features of Alzheimer's disease are used to assess the functional status of this network hub. This allows for a low-cost, non-invasive alternative to FDG-PET to predict the metabolic state of the posterior cingulate cortex of a target individual. By utilizing the metabolic state of the posterior cingulate cortex, the target individual can be efficiently classified into type I or type II based on their glucose metabolism level. Since the glucose metabolism level of the posterior cingulate cortex is closely related to the therapeutic effect of drugs used to treat memory and cognitive impairment, this method can efficiently stratify patients before drug treatment, improving classification efficiency and enabling precision medicine.

[0082] In some embodiments, the clinical data further includes object attribute data corresponding to the target object identifier in at least one object attribute; inputting the memory and cognitive assessment results and the structural data of the target brain region into a pre-trained glucose uptake rate determination model, and outputting the glucose uptake rate of the posterior cingulate cortex corresponding to the target object identifier, includes: inputting the memory and cognitive assessment results, the structural data of the target brain region, and the object attribute data into a pre-trained glucose uptake rate determination model, and outputting the glucose uptake rate of the posterior cingulate cortex corresponding to the target object identifier; the glucose uptake rate is obtained by the pre-trained glucose uptake rate determination model processing the memory and cognitive assessment results, the structural data of the target brain region, and the object attribute data according to the weight information corresponding to different clinical data; the weight information represents the relative importance of the corresponding clinical data to the glucose metabolism analysis of the posterior cingulate cortex.

[0083] Clinical data also includes object attribute data corresponding to the target object identifier in at least one object attribute.

[0084] Among them, object attributes refer to the various attributes or characteristics used to describe and characterize the features of an object.

[0085] Object attribute data refers to the specific descriptions and data about the attributes of an object. It is a detailed description of the object's attributes and may exist in various forms, such as text descriptions, numerical values, labels, vectors, etc. Object attribute data can provide more precise information to facilitate the classification and comparison of objects.

[0086] In practice, when inputting the memory and cognitive assessment results and the structural data of the target brain region into a pre-trained glucose uptake rate determination model, and outputting the glucose uptake rate of the posterior cingulate cortex corresponding to the target object identifier, the memory and cognitive assessment results, the structural data of the target brain region, and the object attribute data can be input into the pre-trained glucose uptake rate determination model. The pre-trained glucose uptake rate determination model can process the memory and cognitive assessment results, the structural data of the target brain region, and the object attribute data according to the weight information corresponding to different clinical data, and obtain the glucose uptake rate of the posterior cingulate cortex corresponding to the target object identifier. Among them, the weight information represents the relative importance of the corresponding clinical data to the glucose metabolism analysis of the posterior cingulate cortex.

[0087] In this embodiment, the clinical data also includes object attribute data corresponding to the target object identifier in at least one object attribute. The glucose uptake rate is determined by inputting the memory and cognitive assessment results, the structural data of the target brain region, and the object attribute data into a pre-trained glucose uptake rate determination model, and outputting the glucose uptake rate of the posterior cingulate cortex corresponding to the target object identifier. The glucose uptake rate is obtained by the pre-trained glucose uptake rate determination model by processing the memory and cognitive assessment results, the structural data of the target brain region, and the object attribute data according to the weight information corresponding to different clinical data. The weight information represents the relative importance of the corresponding clinical data to the glucose metabolism analysis of the posterior cingulate cortex.

[0088] Thus, by combining the memory and cognitive assessment results corresponding to the target object identifier, the structural data of the target brain region, and the object attribute data, the model can accurately predict the glucose uptake rate of the posterior cingulate cortex corresponding to the target object identifier. Furthermore, the model can assign weights to the relative importance of glucose metabolism analysis in the posterior cingulate cortex based on various clinical data, effectively integrating multidimensional clinical data to improve the reliability and accuracy of glucose metabolism prediction. This also expands the application scope of clinical data, allowing for the acquisition of key brain region glucose metabolism characteristics without relying on complex detection methods. This provides more comprehensive and efficient technical support for Alzheimer's disease-related pathological analysis, efficacy evaluation, and patient stratification.

[0089] In some embodiments, classifying a target object into a first type or a second type based on the glucose uptake rate of the posterior cingulate cortex includes: obtaining a preset glucose metabolism threshold; the glucose metabolism threshold is a cutoff value determined based on the Youden index by performing ROC curve analysis on a pre-trained glucose uptake rate determination model; classifying the target object into the first type if the glucose uptake rate of the posterior cingulate cortex meets the glucose metabolism threshold; and classifying the target object into the second type if the glucose uptake rate of the posterior cingulate cortex does not meet the glucose metabolism threshold.

[0090] In practice, when classifying a target object into a first or second type based on the glucose uptake rate of the posterior cingulate cortex, the computer device can first obtain a preset glucose metabolism threshold. This threshold is a cutoff value determined based on the Youden index through ROC curve (Receiver Operating Characteristic Curve) analysis of a pre-trained glucose uptake rate determination model. The glucose uptake rate of the posterior cingulate cortex corresponding to the target object is compared with the preset glucose metabolism threshold. If the glucose uptake rate of the posterior cingulate cortex meets the threshold (i.e., the glucose uptake rate of the posterior cingulate cortex is greater than or equal to the threshold), the target object is classified as the first type; if the glucose uptake rate of the posterior cingulate cortex does not meet the threshold (i.e., the glucose uptake rate of the posterior cingulate cortex is less than the threshold), the target object is classified as the second type.

[0091] The technical solution of this embodiment obtains a preset glucose metabolism threshold; the glucose metabolism threshold is a cutoff value determined by ROC curve analysis of a pre-trained glucose uptake rate determination model based on the Youden index; when the glucose uptake rate in the posterior cingulate cortex meets the glucose metabolism threshold, the target object is identified as a first type; when the glucose uptake rate in the posterior cingulate cortex does not meet the glucose metabolism threshold, the target object is identified as a second type.

[0092] Thus, by classifying target objects based on the glucose metabolism threshold determined by ROC curves and Youden index, it is possible to objectively and accurately segment different groups with varying glucose metabolism levels in the posterior cingulate cortex, thereby achieving stratification of individuals.

[0093] In another embodiment, such as Figure 2 The diagram illustrates a flowchart of a clinical data processing method, including the following steps:

[0094] Step S202: Obtain clinical data corresponding to the target object identifier, including memory and cognitive assessment results, structural data of the target brain region, and object attribute data corresponding to at least one object attribute.

[0095] Step S204: Input the memory and cognitive assessment results, structural data of the target brain region, and object attribute data into the pre-trained glucose uptake rate determination model, and output the glucose uptake rate of the posterior cingulate cortex corresponding to the target object identifier.

[0096] Step S206: Obtain the preset glucose metabolism threshold; the glucose metabolism threshold is the cutoff value determined by ROC curve analysis of the pre-trained glucose uptake rate determination model based on the Youden index.

[0097] In step S208, if the glucose uptake rate in the posterior cingulate cortex meets the glucose metabolism threshold, the target object is classified as type 1; if the glucose uptake rate in the posterior cingulate cortex does not meet the glucose metabolism threshold, the target object is classified as type 2.

[0098] It should be noted that the specific limitations of the above steps can be found in the specific limitations of a clinical data processing method described above.

[0099] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps. It is understood that the steps in different embodiments can be freely combined as needed, and all non-contradictory solutions formed by such combinations are within the scope of protection of this application.

[0100] It should be noted that the relationship between glucose uptake rate in the posterior cingulate cortex, MemTrax assessment results, and MTA score in the medial temporal lobe—that is, the construction of the pre-trained glucose uptake rate determination model—and the differences in lencanezumab treatment efficacy at different glucose metabolism levels, were all determined through experimental analysis. The experimental process will be described below:

[0101] This study explores whether the digital cognitive tool MemTrax combined with MTA can predict PCC metabolic levels and stratify the efficacy response of individuals receiving lencanezumab treatment.

[0102] Methods: Three cohorts of patients with positive amyloid-PET and a Clinical Dementia Rating Scale (CDR) total score of 0.5 or 1.0 were analyzed: two cross-sectional cohorts were used for model construction and internal validation (cohort 1, N=132; cohort 2, N=55), and an independent prospective cohort receiving lencanezumab treatment (cohort 3, N=41, followed for 9 months). MemTrax data, MTA scores, and posterior cingulate cortex glucose metabolism (SUVR) (standardized uptake value ratio) were collected during the study. A linear regression model (i.e., a pre-trained glucose uptake determination model) was built using the cross-sectional data to predict posterior cingulate cortex SUVR using MemTrax accuracy percentage (MTx-%C) combined with MTA, and the data were adjusted for subject attributes such as age and sex. An optimal SUVR cutoff value (a pre-defined glucose metabolism threshold) was determined by ROC analysis to classify the predicted metabolic status into high-metabolic and low-metabolic groups. The model was stratified by metabolic level using data from a prospective cohort, and longitudinal cognitive trajectory changes were compared among individuals grouped according to their baseline glucose metabolism levels.

[0103] Results: Among all brain regions analyzed, posterior cingulate cortex (PCC) metabolism showed the strongest correlation with MemTrax parameters, and the predictive model was significant. In the prospective cohort receiving lencanezumab, baseline predicted PCC metabolism was significantly correlated with ADAS-Cog11 (AD Assessment Scale - Cognitive Subscale) and word recall scores. Longitudinal analysis showed a significant time-group interaction in word recall (p=0.027), with the group with higher predicted metabolic levels exhibiting better trajectories. No significant interactions were found with other cognitive indicators.

[0104] Conclusion: The model constructed using the MemTrax index MTx-%C combined with the medial temporal lobe atrophy index MTA can effectively predict the metabolic status of PCC. Based on this predicted metabolic level, stratified analysis of patients can identify different episodic memory trajectories during lencanezumab treatment, suggesting that this model can serve as a scalable screening tool for patient selection and prognostic stratification in disease-modifying therapy.

[0105] Digital cognitive assessment tools, exemplified by the MemTrax memory test, offer an effective solution for bridging the gap between translational research and clinical application. This test specifically targets episodic memory—the cognitive domain most closely associated with the functional integrity of the precognitive cognitive disorder (PCC). By quantifying key indicators such as recognition accuracy (MTx-%C), reaction speed (MTx-RT), and overall score (MTx-Cp), MemTrax achieves refined, process-oriented cognitive assessment. Its extremely short (<2 minutes), fully automated operation, and use of culturally neutral pictorial stimuli make it highly scalable and applicable to a wide range of clinical practice scenarios.

[0106] With the advent of disease-modifying therapies, the clinical need for convenient and accurate cognitive assessment tools is becoming increasingly urgent. Monoclonal antibodies, such as lencanezumab, have been proven to effectively delay cognitive decline in early-stage Alzheimer's disease (AD). However, individual variability in clinical response remains a significant challenge—even with amyloid clearance, some patients experience only minimal cognitive decline. One of the core goals of precision medicine is to identify reliable biomarkers to screen for patient groups who may benefit more from treatment.

[0107] The functional integrity of the PCC at baseline is expected to serve as a stratified biomarker for this type of disease. Specifically, in the progression of Tau pathology, excitatory synaptic genes in the PCC region are upregulated, and this upregulation is associated with the preservation of cognitive function. This suggests that the PCC may possess pathologically induced plasticity—potentially determining an individual's potential for functional recovery after therapeutic intervention.

[0108] Based on the crucial role of PCC in episodic memory, the following viewpoints are further proposed: MemTrax combined with medial temporal atrophy (MTA) scores can serve as a convenient digital alternative indicator for assessing the integrity of PCC. This study aims to: (1) establish and validate the association between MemTrax performance and MTA with in vivo PCCSUVR; (2) construct a predictive model based on this association and apply it to a prospective cohort receiving lencanezumab treatment to evaluate its stratification value in predicting different cognitive efficacy trajectories. The specific content is as follows:

[0109] method:

[0110] 1. Study Design and Subjects

[0111] 1.1. Research Design

[0112] This study employed a two-stage design, comprising a cross-sectional analysis for model development and a prospective cohort study for model application. The first stage involved a retrospective cross-sectional analysis of enrolled patients to establish the association between the digit cognitive index (MemTrax) and PCCSUVR, thereby creating a predictive model. The subsequent prospective stage applied this model to an independent prospective cohort receiving lencanezumab treatment, assessing its ability to stratify patients based on predicted metabolic status and evaluating the differential trajectory of cognitive efficacy changes over a 9-month period.

[0113] 1.2. Participant Sources:

[0114] The initial screening pool consisted of patients who had undergone comprehensive evaluation, including detailed clinical examination, a combination of cognitive tests, and neuroimaging studies (FDG-PET and AV45-PET). All participants completed a standardized combination of cognitive tests, including the Mini-Mental State Examination (MMSE), the Montreal Cognitive Assessment (MoCA), the Clinical Dementia Rating Scale (CDR), Memtrax, and the Functional Activities Questionnaire (FAQ). The final study cohorts (Group 1 and Group 2) were established using the following key inclusion criteria: (1) positive amyloid protein, confirmed by PET-CT; and (2) a CDR score of 0.5 or 1. Key exclusion criteria included the presence of a major vascular cause, severe mental disorder, or other major neurological disease. This study protocol was approved by the hospital ethics committee, and all participants signed informed consent forms.

[0115] 1.3. Composition of the expected verification queue (Group 3)

[0116] The prospective cohort (Group 3) had the same patient pool as the cross-sectional study, but longitudinal validation recruitment was conducted using an independent protocol. Individuals meeting the following criteria were selected from the cohort of patients who met the core inclusion criteria: (1) amyloid-positive; (2) CDR score of 0.5 or 1; (3) no major vascular cause and Fazekas score <1; (4) no significant mental disorders; and (5) fewer than 5 cerebral microbleeds. These patients subsequently started lencanezumab treatment and signed supplemental informed consent forms for prospective follow-up, and were ultimately included in Group 3. All subjects in this cohort completed comprehensive cognitive assessments, including MMSE, MoCA, CDR, Memtrax, FAQ, and the AD Assessment Scale-Cognitive Subscale (ADAS-Cog11). This group of patients was specifically included to validate predictive models and assess cognitive trajectories during treatment.

[0117] 1.4. Clinical and Neuroradiology Grouping

[0118] The subjects in Group 1 were categorized using two different grouping strategies in order to facilitate subsequent statistical comparisons.

[0119] Clinical diagnosis grouping: Based on the CDR score, the subjects were divided into two clinical diagnosis groups: those with a CDR score of 0.5 were classified as mild cognitive impairment (MCI), and those with a CDR score of 1 were classified as mild AD.

[0120] Neuroradiological grouping: Based on qualitative assessment from clinical FDG-PET reports, subjects were divided into a posterior cingulate cortex (PCC+) group and a posterior cingulate cortex (PCC-) group. This assessment was performed by a certified neuroradiologist unaware of the study hypothesis. Grouping was based on the presence or absence of reduced metabolism in the PCC.

[0121] In subsequent analyses, both grouping methods were used independently to assess inter-group differences in cognitive and radiological indicators.

[0122] 2. Cognitive and Clinical Assessment

[0123] 2.1. ADAS-Cog11

[0124] This study used ADAS-Cog11 (hereinafter referred to as ADAS-Cog) because it is the most widely used version of this scale. ADAS-Cog assesses overall cognitive change through 11 domain-specific items: 7 performance tests (word recall, object naming, instruction execution, constructive practices, conceptual practices, orientation, and word recognition) and 4 assessor-based scales (memorizing test instructions, verbal ability, word lookup ability, and language comprehension). Lower scores on each subtest / subscale indicate better performance, while positive changes from baseline indicate a decline.

[0125] 2.2. MemTrax

[0126] This study was conducted on a desktop / laptop computer using the MemTrax online platform. In short, each MemTrax test displayed a sequence of 50 images: 25 new images and 25 repeating images. Each image was displayed for 3 seconds or until a behavioral response (touching the screen or pressing the spacebar) was initiated. Users were instructed to respond as quickly as possible by touching the screen or pressing the spacebar when a repeating image appeared. At the end of the test, the program calculated and displayed MTx-RT and MTx-%C. The MemTrax composite score (MTx-Cp) was obtained by multiplying the MTx-%C value by the reciprocal of MTx-RT.

[0127] 3. Neuroimaging Acquisition and Processing

[0128] 3.1. FDG-PET Collection

[0129] Before administering the tracer, patients must fast for at least 6 hours and have a confirmed blood glucose level below 10 mmol / L. The imaging protocol includes an intravenous injection of 0.15 MBq / kg. Then a 90-minute acquisition period is performed, followed by PET / CT scan images acquired for the specified duration.

[0130] 3.2. Image Preprocessing and Analysis

[0131] Quantitative analysis of glucose metabolism in local brain regions was performed using Cortex ID Suite. The software automatically extracted mean FDG uptake values ​​from the cerebellum and pons as reference regions, covering 12 bilateral supratentorial regions of interest, including the lateral and medial prefrontal cortex, anterior and posterior cingulate cortex, sensorimotor cortex, precuneus, superior and inferior parietal lobe, lateral and medial temporal lobe, lateral occipital lobe, and primary visual cortex. Individual patient data were compared with an internal dataset of 165 healthy controls. Results were presented in both Z-score and standardized uptake ratio (SUVR) formats, and visualized using a three-dimensional stereoscopic surface projection color-coded map with adjustable color gradation.

[0132] 3.3 MRI Acquisition

[0133] T1-weighted, T2-weighted, and fluid attenuation inversion recovery (T2-weighted inversion) images were acquired on a magnetic resonance scanner.

[0134] 3.4 Medial temporal lobe atrophy

[0135] The MTA scale (used to measure choroidal thickness) is a 5-point scale ranging from 0 to 4. The measurement is based on visual analysis of the choroidal fissure width, temporal horn, and hippocampal height. All MTA measurements are performed visually by a specialist physician.

[0136] 4. Predictive Model Development

[0137] 4.1. Association Analysis

[0138] The association between cognitive parameters and brain glucose metabolism (measured using SUVR) was assessed using correlation analysis, with the specific methodology determined based on the data distribution and research objectives for each cohort. All analyses were performed using R software.

[0139] Specifically, in the first group, the data follows a normal distribution; therefore, the Pearson correlation coefficient is used to assess the linear relationship between the MemTrax parameters and the regional SUVR. In the second group, since the data deviates from a normal distribution, the Spearman rank correlation coefficient is used to assess the monotonic relationship between the ADAS-Cog parameters (total score, word recall, word recognition) and the regional SUVR.

[0140] 4.2. Linear Regression Model

[0141] A multiple linear regression model was constructed to predict glucose metabolism in the right posterior cingulate cortex (PCC_R SUVR). The model was developed using a retrospective untreated cohort (Group 1), with PCC_R SUVR as the dependent variable. The primary independent variables were MTx-%C and MTA, with age and sex included as covariates to control for potential confounding effects. The model is as follows:

[0142]

[0143] This analysis was performed using R software. The model fit was evaluated using R-squared (R²) values, and the statistical significance of each predictor was set at p < 0.05.

[0144] 4.3. ROC Analysis

[0145] Receiver operating characteristic (ROC) analysis was performed on the data from Group 1 using the R package to determine the optimal SUVR cutoff value for posterior cingulate integrity classification. The cutoff value of 0.93 was determined by maximizing the Youden index (J), which optimizes the combined sensitivity and specificity of the classification; this cutoff value was determined before being applied to Group 3.

[0146] 4.4. Linear Mixed Models (LMM)

[0147] Statistical analysis was performed using a linear mixed-effects model (LMM) fitted with the lme4 package in R software. Longitudinal cognitive trajectory analysis employed LMM, with fixed effects including time, group, and their interactions. A random intercept was used to adjust for intra-individual correlations. Despite data loss in group 3 at later time points, LMM was still used to maximize data utility.

[0148] 4.5. General Statistical Standards

[0149] Intergroup comparisons of continuous variables were performed using parametric t-tests in professional statistical analysis and scientific graphing software. A two-tailed p-value less than 0.05 was considered statistically significant.

[0150] result:

[0151] Table 1. Baseline characteristics of retrospective studies (Groups 1-2)

[0152]

[0153] The first set of data was randomly divided into a training set (70%, n=96) and an internal test set (30%, n=38) using stratified random sampling for model development.

[0154] A retrospective study included 187 participants (Group 1, N=132; Group 2, N=55). A subcohort was prospectively defined and followed up from Group 3 (Group 3, N=41) for 9 months.

[0155] Baseline demographic and clinical characteristics of all participants are summarized in Table 1. Groups were well-matched in terms of age, sex, and education level. Baseline cognitive characteristics (including MMSE, MoCA scores, and MemTrax indices) were comparable between Group 3 and the cross-sectional group. Furthermore, baseline PCC_R values ​​remained stable across all groups: 0.91 ± 0.10 for both Group 1 and Group 2.

[0156] Table 2. Longitudinal demographic, cognitive, and metabolic characteristics of the lencanezumab treatment group (Group 3).

[0157]

[0158] The ongoing follow-up has led to a decrease in sample size; some patients have not yet completed the 9-month follow-up.

[0159] The longitudinal data from the three prospective groups primarily focused on the trajectory of changes in cognitive function and biomarkers. The ADAS-Cog score remained stable throughout the study period, with a baseline value of 12.75 ± 6.81 (N=41), 12.10 ± 7.08 (N=26) at 3 months, 13.43 ± 8.03 (N=21) at 6 months, and 13.83 ± 11.25 (N=13) at 9 months. Meanwhile, digital biomarkers such as MTx-%C, MTx-RT, and MTx-Cp all showed a continuous improvement trend from baseline to the 9-month follow-up.

[0160] The grouping criteria for this study required meeting all of the following conditions: positive amyloid protein, CDR score of 0.5 or 1, and exclusion of cognitive impairment caused by vascular cognitive impairment, mental disorders, and other diseases. After this screening, two cross-sectional cohorts were determined: Group 1 (N=132) and Group 2 (N=55).

[0161] This study recruited a prospective cohort of participants receiving lencanezumab treatment, followed by selection according to criterion 2, ultimately forming Group 3, which included 41 participants at baseline. Of these, 26 completed the 3-month follow-up (N=26 for Group 3). At the 6-month follow-up, 21 participants in this group completed the assessment, while 5 had not yet reached their scheduled visit. At the 9-month follow-up, another 13 participants in Group 3 completed the assessment, with 8 still not reaching their scheduled visit.

[0162] To facilitate understanding by those skilled in the art, Figures 3(A)–3(D) illustrate the association between posterior cingulate cortex metabolism and cognitive function and structural integrity. Figure 3(A) shows the association between MemTrax parameters (MTx-%C, MTx-Cp, MTx-RT) and MTA and PCC_R SUVR. Among all brain regions analyzed, PCC_R showed the strongest correlation with all three MTx indices. Figure 3(B) shows the correlation heatmap between PCC_R SUVR and overall cognition and specific cognitive domain scores (ADAS-Cog, word recall, word recognition). PCC_R showed a strong negative correlation with all three cognitive measures. Figure 3(C) shows a box plot comparing PCC_R SUVR in subjects with impaired and unimpaired PCC function in stratified neuroradiological reports. Figure 3(D) shows a box plot comparing PCC_R SUVR based on clinical diagnosis in patients with MCI and mild AD.

[0163] The relationship between PCC_R glucose metabolism and cognitive function:

[0164] As shown in Figure 3(A), Spearman correlation analysis revealed a significant association between MemTrax parameters and SUVR in brain regions. Among all analyzed regions, PCC_R consistently showed the strongest correlation with MemTrax-derived cognitive indices. Specifically, PCC_R metabolism was positively correlated with MTx-Cp (r=0.32, p<0.001), showed the strongest correlation with MTx-%C (r=0.51, p<0.001), and was significantly negatively correlated with MTx-RT (r=-0.34, p<0.001). Notably, among the three MTx measurements, the correlation coefficient between PCC_R SUVR and MTx-%C was the highest across all brain regions. Furthermore, PCC_R metabolism was also negatively correlated with MTA (r=-0.41, p<0.001).

[0165] Consistent with these findings, further analysis of specific cognitive domains highlighted the significant role of the PCC_R. As shown in Figure 3(B), this region exhibited the strongest negative correlation with ADAS-Cog scores (r=-0.58, p<0.001), and also showed significant negative correlations with word recall (r=-0.49, p<0.001) and word recognition (r=-0.28, p<0.05). These results collectively demonstrate that the PCC_R outperforms all other brain regions in its association with overall cognitive function and episodic memory performance.

[0166] PCC_R metabolism and intergroup differences summarized:

[0167] Intergroup comparisons based on neuroimaging assessment and clinical diagnosis revealed significant metabolic alterations in PCC_R. As shown in Figure 3(C), when subjects were divided into PCC lesion and non-lesion groups according to neuroimaging reports, the PCC_R SUVR value was significantly lower in the lesion group (p<0.001). Furthermore, data categorized by clinical diagnosis (Figure 3(D)) showed that PCC_R glucose metabolism was significantly reduced in patients with mild AD compared to patients with MCI (p<0.001).

[0168] These findings confirm a stable and significant association between glucose metabolism in the PCC_R and cognitive function, particularly in episodic memory. Strong correlations observed in multiple cognitive tests, combined with significant metabolic reductions in imaging and clinically defined impairment groups, and a significant negative correlation with medial temporal atrophy (MTA), strongly suggest that the PCC_R is a key neural basis for episodic memory processes. Based on this evidence, low PCC_R metabolism can serve as a biomarker reflecting both cognitive function (cognition) and structural (atrophy) dimensions. Therefore, FDG SUVR may become a sensitive and multidimensional biomarker for assessing the progression of the AD continuum.

[0169] To facilitate understanding by those skilled in the art, Figures 4(A)–4(E) illustrate the development and application of the posterior cingulate cortex glucose metabolism level prediction model (i.e., the pre-trained glucose uptake determination model). Figures 4(A) and 4(B) predict PCC_R SUVR by establishing a linear regression model that includes MTx-%C and MTA scores (adjusted for age and sex). Figure 4(A) shows the significant correlation in the training set (70% of data; R²=0.401), and Figure 4(B) validates the model in the retained test set (30% of data; R²=0.387). Figure 4(C) shows the relative importance of the predictors in the model, indicating that the MTA score is the most contributing factor. Figure 4(D) shows the receiver operating characteristic (ROC) curve analysis of the predicted PCC_R value (Predict_Score) generated by the model, yielding an optimal diagnostic cutoff value of 0.93 (sensitivity=0.82, specificity=0.83). Figure 4(E) shows how all subjects were divided into high metabolic prediction group and low metabolic prediction group based on a cutoff value of 0.93 for subsequent clinical analysis.

[0170] A multimodal predictive model of posterior cingulate cortex metabolic levels: linking cognitive function and structural integrity to metabolic function.

[0171] Based on the established association between PCC and episodic memory, a model for predicting its metabolic activity was developed and validated. First, a linear regression model was used to assess whether MTx-%C and MTA scores, adjusted for age and sex, could jointly predict metabolic impairment in FDG-PCC_R. Using 70% of the data as the training set (Figure 4(A)), the model was statistically significant (p<0.001) and explained a significant proportion of the variation in PCC_R metabolism (R²=0.401). The model was then validated on the remaining 30% of the data (Figure 4(B)), where its predictive performance remained robust (R²=0.387), confirming the model's robustness.

[0172] The contribution of each predictor to the model R² was quantified by relative weight analysis (Figure 4(C)). The results showed that the MTA score explained about 46.1% of the variation in PCC_R metabolism, while MTx-%C explained about 40.5% of the variation.

[0173] To translate this validated predictive relationship into a clinically applicable tool, ROC curve analysis was performed on the predicted PCC_R value (Predict_Score) derived from the model. This analysis aimed to determine the critical value that distinguishes between impaired and preserved PCC integrity. As shown in Figure 4(D), the optimal SUVR cutoff value for Predict_Score, determined by the Youden index, is 0.93. This threshold exhibits a sensitivity of 0.82 and a specificity of 0.83.

[0174] Finally, this threshold was applied to the continuous predictive metabolic score (Predict_Score) of all participants, dividing them into a high metabolic prediction group (Predict_Score ≥ 0.93) and a low metabolic prediction group (Predict_Score < 0.93), as shown in Figure 4(E). This stratification method was reflected in subsequent analyses, enabling the comparison of cognitive and clinical characteristics between individuals with intact and impaired metabolic function prediction in PCC, based on their overall cognitive and structural status.

[0175] Figures 5(A)–5(C) demonstrate the validation of the treatment response assessment prediction score, used to validate the correlation between baseline predicted scores and cognitive measures in patients treated with lencanezumab. Figure 5(A) shows a scatter plot illustrating the correlation between baseline Predict_Score and ADAS-Cog total score. Figure 5(B) shows a scatter plot illustrating the correlation between baseline predicted score and word recall score. Figure 5(C) shows a scatter plot illustrating the correlation between baseline predicted score and word re-recognition score.

[0176] Validation of predictive scores for assessing treatment response:

[0177] To assess the potential utility of the PCC_R Predict_Score in evaluating treatment response, a predictive model derived from the retrospective Group 1 cohort was applied to baseline data from the prospective Group 3 cohort receiving lencanezumab. The correlation between this baseline Predict_Score and cognitive measures, including the commonly used ADAS-Cog scale and its cognitive domains, was then analyzed.

[0178] Figures 5(A)–5(C) show a significant negative correlation between baseline predicted scores and baseline ADAS-Cog total scores (r = -0.6647, p < 0.001). Significant correlations were also observed with key ADAS-Cog sub-scores (including the Word recall test) (r = -0.5457, p < 0.001). These results indicate a meaningful association between predicted scores and the standard cognitive assessment tool ADAS-Cog. The consistent correlation pattern observed in this independent validation cohort suggests that predicted scores may serve as a potential indicator for monitoring the efficacy of lecanemab treatment in patients.

[0179] Predictive validation: Baseline metabolic stratification identifies differential treatment trajectories.

[0180] To further validate the predictive utility of the model, the predicted PCC_R metabolism (Predict_Score) derived from baseline MTx-%C was used to divide the prospective cohort of lencanezumab into a high-metabolic group and a low-metabolic group. Subsequently, during follow-up periods of 3, 6, and 9 months, the longitudinal trajectories of the two groups in six key scores—ADAS-Cog, word recall, word recognition, MTx-%C, MTx-RT, and MTx-Cp—were compared.

[0181] Figure 6 The study revealed distinct cognitive trajectories across the groups. Specifically, the high-metabolic group exhibited a superior trend compared to the low-metabolic group, with lower or more stable ADAS-Cog scores and better memory performance (word recall and recognition). Similarly, MemTrax parameters (MTx-%C, MTx-RT, MTx-Cp) showed differential patterns of change consistent with baseline stratification. These findings suggest that metabolic classification based on a single baseline cognitive indicator (MTx-%C) can identify patients with different cognitive trajectories during lencanezumab treatment.

[0182] The predictive power of metabolic stratification indicators exhibits cognitive domain specificity, primarily evident in the word recall task. Analysis of the time-group interaction showed that word recall (β=0.172, MCI: 0.025 to 0.310, p=0.025) had a statistically significant effect (see Table 3). Table 3 provides the time-group interaction and longitudinal slope analysis of cognitive and metabolic outcomes.

[0183] Table 3

[0184]

[0185] This significant interaction demonstrates a clear difference in cognitive trajectory prediction between metabolomes over time. For the word recall metric, lower scores indicate better performance. Crucially, the high-metabolic group showed a significant performance improvement (slope = -0.145, with a corresponding decrease in error scores), while the low-metabolic group showed a deteriorating trend (slope = +0.027, indicating increased errors). This contrasting pattern suggests that individuals with higher predicted PPC metabolism not only have better baseline function but also exhibit a favorable trajectory of cognitive improvement over time. In contrast, the low-metabolic group lacks this protective advantage and may be in the early stages of cognitive decline. This finding further confirms the model's predictive value in distinguishing individual domain-specific cognitive evolution trajectories. For the remaining results (including word recognition, MTx-%, ADAS-Cog, MTx-Cp, and MTx-RT), the analysis did not show statistically significant differences in longitudinal cognitive trajectory prediction between metabolomes.

[0186] Overall, these longitudinal analyses demonstrate that metabolic stratification based on predictive models has specific and valuable prognostic utility for episodic memory, as demonstrated by the word recall task. Significant differences in memory change trajectories—improvement in the high-metabolic group and decline in the low-metabolic group—strongly validate the model's sensitivity to biologically-based, domain-specific cognitive changes. While the model's predictive value does not generalize across all cognitive domains and metabolic indices, its specific and robust associations with core episodic memory measures confirm its potential relevance in patient stratification and predicting individual memory change trajectories in clinical trials.

[0187] In summary, this study established a robust association between multimodal digital structure surrogate indices and PCC glucose metabolism levels through a two-step validation process. A predictive model integrating MemTrax performance (MTx-%C) and MTA scores was developed and validated, successfully identifying patient subgroups with different episodic memory trajectories in a prospective cohort receiving lencanezumab treatment. This work demonstrates for the first time that a simple digital cognitive index combined with imaging structural biomarkers can serve as a convenient surrogate for assessing metabolic impairment in PCC, providing a practical tool for predicting treatment response in the era of disease-modifying therapies. The integration of cognitive and structural measurements aligns with the co-occurrence of functional and pathological changes in early AD, providing a more comprehensive pathophysiological basis for this model.

[0188] The strong predictive relationship between MemTrax (MTx-%C) and PCC glucose metabolism levels is theoretically based on the well-defined neural structures governing episodic memory. First, MemTrax provides a validated digit detection method specifically designed to assess the encoding phase of episodic memory—a cognitive domain highly dependent on the functional integrity of the default mode network (DMN), in which the PCC acts as a central hub. Therefore, MemTrax results can, to some extent, reflect the functional state of the PCC and DMN.

[0189] Secondly, the MTA score provides crucial complementary structural information, which is essential to the current mainstream theory regarding the origin of PCC dysfunction. The hippocampus (whose integrity is assessed by the MTA) is one of the earliest and most severely affected brain regions in AD. Studies have confirmed a direct coupling between PCC activation and hippocampal activation during successful memory encoding and recognition. This connection can be impaired through multiple pathways, leading to PCC dysfunction: the direct pathway is the disruption of white matter fiber tracts.

[0190] Therefore, this study established and validated a practical framework for translating available clinical data into assessment metrics for key neuroimaging biomarkers. The study demonstrated that a model combining the Simplified Digit Cognition Measure (MTx-%C) with the conventional structural imaging index (MTA) can accurately assess the SUVR of the PCC. This marks a significant methodological shift—transforming clinically accessible metrics into reliable surrogate indicators for key but difficult-to-obtain neuroimaging endpoints. This approach goes beyond simple statistical correlation descriptions, providing an operational model for a core pathophysiological concept: assessing the functional status of the PCC by integrating measures of behavioral output efficiency with the structural integrity of its primary input nodes.

[0191] The significance of PCC functional integrity extends beyond its close association with episodic memory; it also makes it a key biomarker within the theoretical framework of neural reserve and compensation. The findings of this study support a "threshold model" for treatment efficacy. Patients in the "low metabolic predictive value" group may have already crossed the metabolic threshold for functional compensation, with synaptic loss in their PCC region being too severe for cognitive function to be salvaged even with Aβ clearance. Conversely, patients in the "high metabolic predictive value" group may retain sufficient synaptic redundancy (i.e., neural reserve), enabling them to effectively translate Aβ clearance into maintenance or improvement of cognitive function. Contemporary theories of network neurodegeneration propose that, under pathological damage, brain network nodes with high neural reserve can maintain function and play a compensatory role, while nodes with low reserves will degenerate. As the hub of the default mode network (DMN), the PCC is a key site of action in this dynamic process. Previous studies have shown that the α-secretase ADAM10, as a response mechanism to the progression of Tau pathology, may participate in this neural reserve-related compensatory process.

[0192] The predictive model in this application, which categorizes patients into high-metabolic and low-metabolic groups using MemTrax, may capture this continuum of compensatory capacity, ranging from preservation to exhaustion. Individuals in the "high-metabolic" group retain higher levels of neural function—their PCCs and connectivity networks maintain the synaptic and functional plasticity required for adaptation. In contrast, the "low-metabolic" group may represent a state of failure or exhaustion of these compensatory mechanisms.

[0193] This framework offers an explanation for clinical findings: while lencanezumab reduces amyloid burden, the conversion of this pathological change into cognitive benefit depends on the brain's residual capacity for functional reorganization. Patients with higher PCC SUVR (high-metabolic group) possess the physiological basis needed to utilize the cleared pathological "space" for network stabilization or recovery, resulting in a superior trajectory of cognitive change. Conversely, exhausted patients (low-metabolic group) lack this adaptive potential, limiting the clinical efficacy of amyloid clearance. This logic aligns with the research findings: the stratification of the model specifically predicted differences in episodic memory, the cognitive domain most closely associated with PCC integrity.

[0194] In addition to incorporating MTx-%C, MTA, age, and sex as predictors, a wider range of digital biomarkers, plasma assays (such as p-Tau217), or other modifiable factors can be included to further improve predictive accuracy and biological specificity.

[0195] Based on the same inventive concept, this application also provides a clinical data processing apparatus for implementing the clinical data processing method described above. The solution provided by this apparatus is similar to the implementation described in the above method; therefore, the specific limitations in one or more clinical data processing apparatus embodiments provided below can be found in the limitations of the clinical data processing method described above, and will not be repeated here.

[0196] In one exemplary embodiment, such as Figure 7 As shown, a clinical data processing apparatus is provided, comprising: an acquisition module 710, a prediction module 720, and a classification module 730, wherein:

[0197] The acquisition module 710 is used to acquire clinical data corresponding to the target object identifier; the clinical data includes memory and cognitive assessment results and structural data of the target brain region; the target brain region is a brain region associated with the pathological characteristics of Alzheimer's disease; the target object identifier is the object identifier corresponding to the target object with memory and cognitive impairment.

[0198] The prediction module 720 is used to input the memory and cognition assessment results and the structural data of the target brain region into a pre-trained glucose uptake rate determination model, and output the glucose uptake rate of the posterior cingulate cortex corresponding to the target object identifier; the pre-trained glucose uptake rate determination model is obtained by performing correlation analysis between the memory and cognition assessment test results corresponding to the subject identifier and the structural test data of the target brain region and different brain regions corresponding to the subject identifier; the subject identifier is the object identifier corresponding to the subject with memory and cognition impairment.

[0199] The classification module 730 is used to classify the target object identifier into a first type or a second type based on the glucose uptake rate of the posterior cingulate cortex; wherein the glucose metabolism level of the posterior cingulate cortex corresponding to the object identifier belonging to the first type is better than the glucose metabolism level of the posterior cingulate cortex corresponding to the object identifier belonging to the second type.

[0200] In one embodiment, the clinical data further includes object attribute data corresponding to the target object identifier in at least one object attribute; the prediction module 720 is specifically used to input the memory and cognition assessment results, the structural data of the target brain region, and the object attribute data into the pre-trained glucose uptake rate determination model, and output the glucose uptake rate of the posterior cingulate cortex corresponding to the target object identifier; the glucose uptake rate is obtained by the pre-trained glucose uptake rate determination model processing the memory and cognition assessment results, the structural data of the target brain region, and the object attribute data according to the weight information corresponding to different clinical data; the weight information represents the relative importance of the corresponding clinical data to the glucose metabolism analysis of the posterior cingulate cortex.

[0201] In one embodiment, the target brain region includes the medial temporal lobe, and the acquisition module 710 is specifically used to acquire structural evaluation data of the medial temporal lobe corresponding to the target object identifier; the structural evaluation data is used to characterize the structural integrity of the medial temporal lobe; and the structural evaluation data of the medial temporal lobe corresponding to the target object identifier is used as the structural data of the target brain region.

[0202] In one embodiment, the acquisition module 710 is specifically used to acquire the atrophy score of the medial temporal lobe corresponding to the target object identifier; and use the atrophy score as the structural evaluation data.

[0203] In one embodiment, the memory and cognition assessment result includes the MemTrax assessment result obtained by performing a memory and cognition assessment based on the MemTrax system; the MemTrax assessment result includes at least the memory and cognition accuracy rate.

[0204] In one embodiment, the classification module 730 is specifically used to obtain a preset glucose metabolism threshold; the glucose metabolism threshold is a cutoff value determined based on the Youden index by performing ROC curve analysis on the pre-trained glucose uptake determination model; if the glucose uptake rate in the posterior cingulate cortex meets the glucose metabolism threshold, the target object is classified as the first type; if the glucose uptake rate in the posterior cingulate cortex does not meet the glucose metabolism threshold, the target object is classified as the second type.

[0205] Each module in the aforementioned clinical data processing device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the operations corresponding to each module.

[0206] In one exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 8 As shown, the computer device includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The input / output interface is used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, Near Field Communication (NFC), or other technologies. When the computer program is executed by the processor, it implements a method for processing clinical data. The display unit is used to form a visually visible image and can be a display screen, projection device, or virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.

[0207] Those skilled in the art will understand that Figure 8 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0208] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0209] In one embodiment, a computer-readable storage medium is provided having a computer program stored thereon that, when executed by a processor, implements the steps in the above method embodiments.

[0210] In one embodiment, a computer program product is provided, including a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0211] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.

[0212] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.

[0213] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.

[0214] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for processing clinical data, characterized in that, The method includes: Acquire clinical data corresponding to the target object identifier; the clinical data includes memory and cognitive assessment results and structural data of the target brain region; the target brain region is a brain region associated with the pathological features of Alzheimer's disease; the target object identifier is the object identifier corresponding to the target object with memory and cognitive impairment; The memory and cognitive assessment results and the structural data of the target brain region are input into a pre-trained glucose uptake rate determination model, which outputs the glucose uptake rate of the posterior cingulate cortex corresponding to the target object identifier. The pre-trained glucose uptake rate determination model is obtained by performing correlation analysis between the memory and cognitive assessment test results corresponding to the subject identifier and the structural test data of the target brain region and different brain regions corresponding to the subject identifier. The subject identifier is the object identifier corresponding to the subject with memory and cognitive impairment. Based on the glucose uptake rate of the posterior cingulate cortex, the target object is classified into a first type or a second type; wherein, the glucose metabolism level of the posterior cingulate cortex corresponding to the object object belonging to the first type is better than the glucose metabolism level of the posterior cingulate cortex corresponding to the object object belonging to the second type.

2. The method according to claim 1, characterized in that, The clinical data also includes object attribute data corresponding to the target object identifier in at least one object attribute; the step of inputting the memory and cognitive assessment results and the structural data of the target brain region into a pre-trained glucose uptake rate determination model, and outputting the glucose uptake rate of the posterior cingulate cortex corresponding to the target object identifier, includes: The memory and cognition assessment results, the structural data of the target brain region, and the object attribute data are input into the pre-trained glucose uptake rate determination model, and the glucose uptake rate of the posterior cingulate cortex corresponding to the target object identifier is output. The glucose uptake rate is obtained by the pre-trained glucose uptake rate determination model, which processes the memory and cognition assessment results, the structural data of the target brain region, and the object attribute data according to the weight information corresponding to different clinical data; the weight information represents the relative importance of the corresponding clinical data to the analysis of glucose metabolism level in the posterior cingulate cortex.

3. The method according to claim 1, characterized in that, The target brain region includes the medial temporal lobe, and the acquisition of clinical data corresponding to the target object identifier includes: Obtain structural evaluation data of the medial temporal lobe corresponding to the target object identifier; the structural evaluation data is used to characterize the structural integrity of the medial temporal lobe. The structural assessment data of the medial temporal lobe corresponding to the target object identifier is used as the structural data of the target brain region.

4. The method according to claim 3, characterized in that, The step of obtaining the structural evaluation data of the medial temporal lobe corresponding to the target object identifier includes: Obtain the atrophy score of the medial temporal lobe corresponding to the target object identifier; The atrophy score is used as the structural assessment data.

5. The method according to claim 1, characterized in that, The memory and cognition assessment results include MemTrax assessment results obtained based on the MemTrax system; the MemTrax assessment results include at least the memory and cognition accuracy rate.

6. The method according to claim 1, characterized in that, The step of classifying the target object into a first type or a second type based on the glucose uptake rate of the posterior cingulate cortex includes: Obtain a preset glucose metabolism threshold; the glucose metabolism threshold is a cutoff value determined based on the Youden index by performing ROC curve analysis on the pre-trained glucose uptake rate determination model. If the glucose uptake rate in the posterior cingulate cortex meets the glucose metabolism threshold, the target object is identified and classified as the first type. If the glucose uptake rate in the posterior cingulate cortex does not meet the glucose metabolism threshold, the target object is identified and classified as the second type.

7. A clinical data processing device, characterized in that, The device includes: The acquisition module is used to acquire clinical data corresponding to the target object identifier; the clinical data includes memory and cognitive assessment results and structural data of the target brain region; the target brain region is a brain region associated with the pathological features of Alzheimer's disease; the target object identifier is the object identifier corresponding to the target object with memory and cognitive impairment; The prediction module is used to input the memory and cognitive assessment results and the structural data of the target brain region into a pre-trained glucose uptake rate determination model, and output the glucose uptake rate of the posterior cingulate cortex corresponding to the target object identifier; the pre-trained glucose uptake rate determination model is obtained by performing correlation analysis between the memory and cognitive assessment test results corresponding to the subject identifier and the structural test data of the target brain region and different brain regions corresponding to the subject identifier; the subject identifier is the object identifier corresponding to the subject with memory and cognitive impairment. The classification module is used to classify the target object identifier into a first type or a second type based on the glucose uptake rate of the posterior cingulate cortex; wherein the glucose metabolism level of the posterior cingulate cortex corresponding to the object identifier belonging to the first type is better than the glucose metabolism level of the posterior cingulate cortex corresponding to the object identifier belonging to the second type.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.

Images