Parkinson’s disease (PD) is the 2nd most common form of neurodegenerative disease after Alzheimer’s disease. It is affecting 1% of the population above 60. The share of the Taiwanese population above 60 is already 20% and increasing, so it is a major health care issue for the society. PD is a neurodegenerative disease caused by the loss of dopaminergic neurons, and patients gradually lose their motor functions and ability to take care of themselves before dying from the disease. In this project we focus on the motor symptoms, mainly tremor, dyskinesia, and bradykinesia, which significantly cause impairment in the quality of life of patients and disability in daily activities.

The most common assessment tool for PD–the Unified Parkinson’s Disease Rating Scale–is only suitable for clinical usage and provide a rather subjective score. Our aim is to automate and improve the quantification of the motor symptoms, first for clinical evaluation and later for self evaluation using a smart phone. We will develop a computer vision system for video based quantification of motor symptoms that utilizes deep learning and convolutional neural networks. The system will analyze micromotions that are not visible to the human eye.

We believe that this could reduce the anxiety and enable individualized adjustment of the medication dosage to better control the symptoms and improve the quality of life of patients with Parkinson’s disease. We will recruit 120 participants (60 PD, 20 pre-diagnosis PD, and 40 age- and sex-matched healthy volunteers) from the Kaohsiung Medical University Hospital or communities for data collection and clinical evaluation of our systems effect on dosing of the medication and life quality. The goal is to assist PD patients to maintain control over their life and reduce the growing health care burden.

Keywords: Parkinson’s disease (PD), assisting technology, clinical diagnosis, quantified self, Artificial intelligence (AI), Artificial neural networks (ANNs), Deep learning, Artificial narrow intelligence (ANI).

微小動作分析技術應用於改善帕金森氏症病患之照護

帕金森氏症是僅次於阿茲海默症第二常見的一種神經退行性疾病，它影響超過1%以上的60歲人口，而台灣60歲以上人口超過20%且逐年增加，因此這也成為台灣社會在健康照護主要的支出之一。帕金森氏症是一種因多巴胺神經細胞減少所導致的神經退化疾病，其患者會逐漸失去運動能力並無法照顧他們自己直到死亡。 在這項計畫中，我們著重在顫抖、動作遲緩和運動困難的運動症狀，這些症狀會嚴重影響患者的日常並造成生活品值下降。目前臨床上最常用的評估工具是UPDRS帕金森氏症評量表，但它只提供相對較為主觀的結果。

我們的目標是自動化並改善這些運動症狀的量化方式，且預計先應用於協助臨床評估並在之後透過智慧型手機達到自我評估。 我們計畫開發以影像分析方法量化運動症狀之電腦視覺系統，也運用深度學習及捲積神經網路等技術，這系統將會分析人眼不易察覺的細微動作。我們相信透過這項技術評估症狀並調整治療藥物劑量可改善患者生活並減少焦慮。我們預計從高雄醫學大學的醫院招募120位參與者（60位確診和20位預診為帕金森氏症之患者、40位年齡及性別相符的健康自願者），蒐集的資料將用於臨床上評估此系統對治療藥物劑量與生活品質的效果。我們的最終目標是協助帕金森氏症患者維持生活所需的控制能力，並減少不斷增加的照護負擔。

關鍵字: 帕金森氏症、輔助科技、臨床診斷、人工智慧、類神經網路、深度學習、弱人工智慧.

Principal Investigator: Ass. Prof. Torbjörn Nordling.

Co-principal Investigators: Ass. Prof. Chi-Lun Lin, Dept. of Mechanical Engineering, National Cheng Kung University.
Ass. Prof. Chun-Hsiang Tan, Dept. of Neurology, Kaohsiung Medical University.

Sponsor: Ministry of Science and Technology in Taiwan 

Grant identifiers: 108-2218-E-006-046, 109-2224-E-006-003.

Duration: 2019-06-01 to 2021-08-31.

Members: Akram Ashyani, Esteban Roman, Jose Chang, YuShan Lin, Jacob Chen, Gavin Vivaldy, Ric Tu, Austin Su, Chin Sheng Lin, Torbjörn Nordling.