By: Eric Wen
Parkinson’s Disease (PD) is one of the fastest-growing neurological diseases in the world. Yet, 20% of individuals with PD remain undiagnosed due to shortages of neurologists and limited access to healthcare in developing countries. An early diagnosis is important for improving the quality of life of PD patients, making fast diagnosis an important component of PD treatment. However, this is rendered difficult due to the current diagnosis standards that require in-person clinics and an expert that assesses symptoms and makes a diagnosis, which may not be present in developing countries. A recent study published by Rahman et al. (2021) investigates if PD can be diagnosed remotely through a web-based audio recording software, reducing the need for in-person clinical meetings.
726 unique participants (36% of whom have received a PD diagnosis) were recruited for this study. They were asked to repeat the popular sentence “the quick brown fox jumps over the lazy dog” and were recorded using a web-based tool called PARK, which can be found on parktest.net. Most participants performed the test unassisted at their homes. After recording all participants, the audio files were cleaned up (by removing background noises) and the pronunciations were extracted. 4 machine learning programs were then used to analyze the data, which identified different vocal impairments that occurred and diagnosed the patient. After accounting for gender and age variations, it was found that the best program had an accuracy score of 74% when diagnosing PD (74% of its diagnoses were correct), while the other 3 had scores between 72-74%.
The PARK tool is a potentially game-changing way of diagnosing PD. This is because it only requires an internet connection, a camera and a microphone to function properly, and can also be accessed by anyone around the world since it is a web-based program. Though the program is currently far from perfect, future development could potentially lead to early diagnoses solely from audio recordings, which could greatly increase treatment success and the quality of life of PD patients. The study also reveals that the diagnostic accuracy of all programs were the same for recordings done under the supervision of a professional and without supervision, suggesting that the test can be done by patients themselves in their own homes. In the future, it could even be developed as a downloadable application that could test for PD speech abnormalities or assist neurologists during virtual meetings.
Though the results of this study are promising, there are still many improvements that need to be made before the program can be used as a fully-fledged diagnostic tool. Firstly, the diagnoses that all programs provide are still not consistently accurate, and the programs should also examine abnormalities other than vocal impairment (such as tremors). In addition, most participants examined in the study were native English speakers, and future research should focus on accurately diagnosing non-native English speakers and even speakers of other different languages. Lastly, many individuals in developing nations also do not have access to the internet, meaning that they would still be unable to access the PARK program. Therefore, future development of the program should emphasize offline data collection and diagnosis, so that the PARK program would be usable with or without internet connection. In the end, the PARK program developed by Rahman et al. represents the next step towards fast and efficient diagnosis of PD, and could be a life-changing program for millions of PD patients around the world (Rahman et al., 2021).
Rahman W, Lee S, Islam MS, Antony VN, Ratnu H, Ali MR, Mamun AA, Wagner E, Jensen-Roberts S, Waddell E, Myers T, Pawlik M, Soto J, Coffey M, Sarkar A, Schneider R, Tarolli C, Lizarraga K, Adams J, Little MA, Dorsey ER, Hoque E
Detecting Parkinson Disease Using a Web-Based Speech Task: Observational Study
J Med Internet Res 2021;23(10):e26305