Babusi Nyoni is an interesting guy. From creating the “world’s first AI football commentator” for UEFA in 2016 to working with the United Nations on a number of artificial intelligence-related projects, it’s fair to call the Zimbabwean an Artificial Intelligence expert.
The same man worked with triple black agency to develop an application that can assist healthcare professionals in diagnosing Parkinson’s disease. Patana AI uses “computer vision to assess posture and gait on mobile devices and uses the device accelerometer to analyse hand tremors.”
Simply put, the app does not detect Parkinson’s disease, rather it allows for the collection of data by healthcare practitioners in order to inform such a prognosis.
Babusi Nyoni
One of the key partners who were involved in the building up of the Patana AI application was Ekuphumuleni Geriatric Nursing Home “research insights and usability testing”.
Data scarcity is still an issue but it’s worth noting that more often than not, the main challenge is discoverability rather than actual absence.
Babusi Nyoni
Patana AI makes use of a custom-trained model for the posture classification and uses an algorithm for gait and tremor assessments. Nyoni explained that the posture model is currently “being retrained every two weeks to improve accuracy”.
We asked Babusi what devices optimally run the Patana AI application and he shared that at the moment devices with the Snapdragon 835 and at least 2GB of RAM get the job done. What does that mean for those you who aren’t concerned with specs? Any top-end and midrange device made in the last two years will probably be able to run the application without too much of a hassle.
It’s great to see such strides being made on our continent and hopefully, the work done by Triple Black Agency and Babusi in bringing Patana AI to fruition inspires other people looking to enter the field of artificial intelligence in the health sector.
Share
Home
8 comments
Forgive my ignorance Farai, but what will be the purpose of this app? From my knowledge of Parkinson’s, by the time one has such a posture, they probably know they have the disease. I’m in no way trying to shoot down the work done here, just trying to understand the use cases of this app.
Some people might know they have the disease but a lot of people might not know and the app isn’t for you to download and use at home. It’s for healthcare professionals to collect data when they are attempting to diagnose if, in fact, the patient has Parkinson’s. I don’t think all cases are as simple as looking at someone with just your eyes and determining “Oh! they have Parkinson’s”.
Thanks, I understand the app is for home use and I understand it’s for data collection. When you say some people might not know they have the disease, I think you are not talking from a point of experience. Posture change is usually the last symptom of the disease. Usually people with the disease start of with simple symptoms like voice changes and handwriting changes. When the symptoms progress to muscle stiffness, that’s when people usually check in to see a doctor. In my opinion, we should be using tech to diagnose from these early symptoms. So my question again, after we have collected this data and we can diagnose the disease accurately from the posture, what then are the use cases for the app? Shouldn’t we be building AI to diagnose from the early symptoms rather than the late symptoms?
*not for home use
Thanks
I think I failed to clearly explain myself in the first response. There are a number of movement diseases that can be confused with Parkinson’s (i.e progressive palsy, MS atrophy) not to mention there are also different types of Parkinson diseases (e.g PT Parkinson, drug-induced Parkinson) so the AI application helps healthcare professionals to collect data and from this data, the hope is that they can narrow down the list of which ailments are actually affecting the patient.
Also, more importantly, “Parkinson’s disease” was in the title of the article because it is what I felt is most relatable. The application is not for Parkinson’s only but more specifically for posture and gait assessment to pick up the number of different diseases which can be diagnoses by looking at a patient’s gait and posture – some of which I mentioned above. The app is not exclusive to Parkinson’s. I hope that clarifies things.
Thanks for the explanation, I now see from your angle. I had assumed the app would just be for Parkinson’s hence I challenged the use cases. I hope Babusi will look into developing something for the early stages of Parkinson’s and move to other diseases like the one you wrote about for Corona.
thanks