Linear predictive coding of EEG Activity for Diagnosis of Parkinson's Disease (LEAD-PD)

需要可靠和有效的工具来1)诊断和区分帕金森病(PD)与其他临床特征相似但预后和治疗不同的运动障碍，2)量化和跟踪PD随时间的运动和认知障碍症状，以及3)评估对治疗变化的反应，以优化症状控制。目前用于这些目的的工具主要包括临床量表和问卷；然而，结果可能是高度可变的。因此，帕金森病患者迫切需要准确可行的生物标志物。我们提出了一种新颖的、神经生理学的机器学习方法来满足这一需求。我们开发了脑电活动的线性预测编码(LPC)来诊断帕金森病(Lead-PD)。铅-帕金森病并不关注频段，而是使用5分钟的休息数据捕捉帕金森病患者功率谱的关键差异。初步结果显示，铅-帕金森病在独立的验证集上达到了85%的敏感性/特异性，超过了其他潜在的帕金森病临床生物标志物。这项拟议研究的总体目标是开发一种新的、客观的生物标志物，用于诊断帕金森病并跟踪其进展和治疗反应。在这个提案中，我们将检验铅-帕金森病的总体假设，即铅-帕金森病包括帕金森病的诊断和临床特征的多样性/严重性。我们的具体目标是：目标1：确定脑电对帕金森病的诊断作用。我们的工作假设是，铅-帕金森病诊断指数将把帕金森病患者与对照组、原发性震颤患者和帕金森综合征患者区分开来，帕金森综合征包括多系统萎缩(MSA)和进行性核上性瘫痪(PSP)，包括阿尔茨海默病相关痴呆(ADRD)，如路易体痴呆(DLB)。目的2：探讨脑电图在预测帕金森病关键症状严重程度和进展中的作用。我们的工作假设是，铅-帕金森病运动能力和认知指数将预测这些关键症状在两年内的基线严重性和纵向恶化。目的3：确定EEG在评估PD患者对DBS治疗的运动反应中的作用。我们的工作可以为PD和阿尔茨海默病(AD)和相关痴呆(ADRD)，如阿尔茨海默病(AD)和路易体痴呆(LBD)，包括帕金森病痴呆(PDD)和DLB，贡献新的生物标记物和实时应用。由于铅-PD指数可用于区分症状严重程度，包括认知障碍和在一连串疾病阶段的严重程度，我们的发现与ADRD(如LBD)直接相关。

Reliable and efficient tools are needed to 1) diagnose and differentiate Parkinson’s disease (PD) from other movement disorders with similar clinical features but with different prognosis and treatment, 2) quantify and track motor and cognitive symptoms of PD over time, and 3) assess response to treatment changes for optimization of symptom control. Current tools for these purposes mainly consist of clinical scales and questionnaires; however, the results can be highly variable. Thus, there is a critical need for accurate and feasible biomarkers in PD. We propose a novel, neurophysiological, machine-learning approach to fulfill this need. We have developed Linear predictive coding (LPC) of EEG Activity for the Diagnosis of PD (LEAD-PD). Rather than focusing on frequency bands, LEAD-PD captures critical differences in the power spectra of PD patients using <5 minutes of resting data. Preliminary results show that LEAD-PD achieves >85% sensitivity/specificity in independent validation sets, surpassing other potential clinical biomarkers for PD. The overall objective of the proposed research is to develop a novel, objective biomarker for diagnosing PD and tracking its progression and response to treatment. In this proposal, we will test the overall hypothesis that that LEAD-PD captures PD diagnosis and diversity/severity of clinical features. Our specific aims are: AIM 1: Determine the diagnostic role of EEG in PD. Our working hypothesis is that the LEAD-PD diagnostic index will distinguish patients with PD from controls, patients with essential tremor, and Parkinson-plus syndromes such as multiple system atrophy (MSA) and progressive supranuclear palsy (PSP), including Alzheimer-related dementias (ADRD) such as Dementia with Lewy Bodies (DLB). AIM 2: Determine the role of EEG in predicting the severity and progression of key symptoms of PD. Our working hypotheses are that the LEAD-PD motor and cognitive indices will predict both baseline severity and longitudinal worsening of these key symptoms over 2 years. AIM 3: Determine the role of EEG in assessing the motor response to DBS treatment in PD. Our work could contribute novel biomarkers and real-time applications for PD and for Alzheimer’s disease and related dementias (ADRD) such as Alzheimer’s dementia (AD) and Lewy Body Dementias (LBD), including Parkinson’s disease dementia (PDD) and DLB.  Because the LEAD-PD index may be utilized to discern symptom severity including cognitive impairment and severity across a continuous spectrum of disease stages in synucleinopathies, our findings are directly related to ADRD such as LBD.