According to a report, Parkinson’s disease(PD) is one of the fastest-growing neurological diseases in the world. However, it is challenging to diagnose as it depends on the symptoms like tremors and slowness and often appears after several years at the onset of the disease.
In recent weeks MIT researchers made a big announcement that they have developed an artificial intelligence model that represents the success in detecting Parkinson’s disease from breathing patterns so that Parkinson’s disease can be detected earlier and contactless using radio waves. In this article, we are going to look at the following points related to this important news.
Table of content
- Need for this model
- How is the model developed?
- Results
- Properties
Need for this model
According to the journal published on natural medicine, currently, no effective markers or biomarkers are available for diagnosing and tracking Parkinson’s disease(PD). Using this model, we can not only diagnose the PD but also track its progress. As discussed above, this model requires breathing patterns for diagnosing and tracking PD, It becomes a competitive biomarker of PD that works contactless.
The history of Parkinson’s disease is worst because data says that over 1 million people in the united states are surviving with PD as of 2020, causing a unique budget of $52 billion per year. The progression of PD cannot be stopped using any drug, and the lack of effective diagnostic biomarkers makes this disease challenging. So the requirement of early diagnosis and progress checking makes this model very important for a better clinical system.
How is the model developed?
Before making an effective diagnosis system for PD, the researchers investigated various traditional systems and biomarkers, among which cerebrospinal fluid, blood biochemical and neuroimaging are also efficient but costly and unsuitable for frequent testing. None of them provides an early diagnosis.
According to James Parkinsons, there is a relationship between breathing and PD. this made the researchers use a large dataset comprising records of 7671 individuals. Internally, the model uses a neural network with an attention mechanism that helps make predictions concerning sleep and electroencephalogram.
The model can learn the auxiliary task of predicting a person’s quantitative electroencephalogram (qEEG) from nocturnal breathing so that overfitting of the model can be avoided in addition to interpreting the model’s outcome. The below image represents the major components of this biomarker.
Image source
By looking at the image we can say that the system can extract breathing patterns from the human body using a belt or radio signals. After extraction, it processes the patterns through a neural network and infers whether the person is diagnosed with PD or not and if yes, assesses the severity in accordance with MDS-UPDRS.
Results
As discussed in the above points, the dataset for training this model had 7671 data points. When talking about the systems detection capabilities, the researchers claim it can detect the PD with an area-under-the-curve of 0.90 and 0.85 on held-out and external test sets. However, when it comes to predicting the severity and progression of PD, this system can give an R-score of 0.94 and a p-value of 3.6 x 10^-25. Used data was separated into two subsets: data came from the breathing belt setting and the wireless setting. The below picture represents the results of the system when detecting PD.
Image source
And the below image represents the results while the system is predicting the severity and progress of Parkinson’s disease.
image source
Properties
Looking at the above points, and the journal, we can say that biomarker has the following properties:
- This single work can accurately identify the PD status and also can predict severity and progression.
- This work removes the high cost and experience required by the traditional ways to diagnose PD.
- This system has the potential for becoming a new biomarker because all the desirable features are added to it.
- Unlike traditional methods and biomarkers, this model is very sensitive to small changes and can track them very easily.
- This system can help in the early diagnosis of PD.
- This system can be used by patients in their homes because it is easy.
Final words
In this article, we looked at an AI-enabled system that uses Artificial Intelligence and machine learning for classification (PD — Yes or No) and regression modelling(severity and progress of PD) together. Since it came out as a great change in the field of medical science, it represents the importance of AI in today’s scenarios. Such use cases are not only available in the health care domain but also in every domain.
We at DSW | Data Science Wizards are constantly working to complete our vision of making AI available for everyone. Our flagship-platform UnifyAI has the ability to be utilised in any domain. We designed its components in such a way that they can be fitted into traditional systems very easily and perform data-related operations faster than older or other components. We ensure that all important data can be utilized in the life cycle of data science to derive data-driven decisions and optimize them for business growth.
About DSW
Data Science Wizards (DSW) is an Artificial Intelligence and Data Science start-up that primarily offers platforms, solutions, and services for making use of data as a strategy through AI and data analytics solutions and consulting services to help enterprises in data-driven decisions.
DSW’s flagship platform UnifyAI is an end-to-end AI-enabled platform for enterprise customers to build, deploy, manage, and publish their AI models. UnifyAI helps you to build your business use case by leveraging AI capabilities and improving analytics outcomes.
Connect us at contact@datasciencewizards.ai and visit us at www.datasciencewizards.ai