Scroll Image

About Parkinson's Disease

Parkinson’s Disease is a movement disorder impacted by the nervous system. It causes neurons in the brain to become damaged and weakened which impacts lifestyle such as movement issues, tremors, stiffness, and impaired balance. Severe PD patients might have troubles with walking, talking, and completing other daily tasks. The damage in the neurons causes the affected individuals to have a decrease in dopamine.

As the modern world has become more digitized, typing and computer usage has become a large part of our lives. People who suffer from PD may have difficulty typing or operating computes due to reduced motor function. In fact, research has implied that certain typing patterns can serve as an indicator of Parkinson's disease risk.

Through the analysis of these patterns, we hope to identify those who may be at risk and provide earlier intervention.

About Our Dataset

Our original dataset was collected by the Massachusetts Institute of Technology (MIT) on 85 subjects (43 with Parkinsons, 42 without) from Madrid, Spain. It contains keystroke logs from subjects both with and without PD. Typing data was collected using standardized equipment across all participants.

Besides the keylogged data (the key pressed, hold time, release time, etc.), finger tapping tests were also taken that measured the speed and rhythm of finger movements, providing additional insights of the likelihood of PD.

One of the main takeaway points of the dataset is the Unified Parkinson's Disease Rating Scale (UPDRS) score. UPDRS is based upon a questionnaire that subjects complete, that assesses subjects on motor and non-motor symptoms, each on a scale from 0-4. This dataset focused specifically on Part 3 of the UPDRS, measuring motor function.

Let's Look At The Data!

data Image

Regressing WPM on UPDRS

Hover over the graph to see details!

  • For both populations, there is a negative trend in the regression line, meaning that as the UPDRS score increases, the typing speed (WPM) decreases.
  • The slope of regression line for the combined population is a decrease of 1.24 WPM for each point increase in UPDRS score. For subjects with PD there is a decrease of 2.50 WPM and for those without PD, it decreases by 4.26 WPM.
  • Subjects without PD have a much wider range of WPM compared to those with PD suggesting that the typing speed of those with PD is more consistent and may be more predictable.
  • There is a clear linear division in the UPDRS between the two groups, which emerges around 6-7 UPDRS. It may be possible to classify an individual’s Parkinson's risk based on their UPDRS score!

Can We Predict Parkinson’s Disease?

predict Image

Density Histogram Comparison of WPM

Hover over the graph to see details!

  • Although it would be expected that there would be a vast difference in WPM since PD impacts one's movement, our data shows that there is NOT a drastic difference in typing speed, as the distributions between subjects with PD and those without appear quite similar.
  • One noticeable difference, however, is the extremities in the data which belong to subjects without Parkinson’s Disease. The highest WPM in the dataset is 257.2 WPM, which is considered exceptionally fast.
  • Due to the similar distributions, it is likely that typing speed alone is not a sufficient metric for diagnosis. We cannot distinguish between those with and without PD, as a subject could belong to either population regardless of their typing speed.

It's Too Difficult To Predict!

Let's See How YOU Compare

Typing Test Instructions

  1. Click "Start typing here…" below to begin the typing test.
  2. Type the provided text as accurately and quickly as possible.
  3. Make sure to hit space in between words.
  4. After completing the test, you will receive feedback on your typing performance.



Scroll down to take the test!

Typing Test

Time: 15s

Conclusion

Our project aimed to explore the relationship between typing patterns and Parkinson's disease risk hoping that with analysis of typing speed, it would enable earlier intervention and symptom management.

Initially, it seemed promising that there was a clear difference in typing speeds between individuals of both populations. However, deeper analysis reveals that typing speed alone is not sufficient enough to diagnose Parkinson’s. In fact, individuals without PD could still receive a high UPDRS score from the typing test.

The distributions of typing speed for both populations are too similar, to the point that there is no definitive threshold that would allow us to confidently distinguish between those with and without disease. Therefore any predictions made should be considered speculative and used only for exploratory purposes, NOT for clinical evaluation.

This emphasizes the importance of considering a myriad of factors other than just typing speed when assessing Parkinson’s risk and severity.

conclusion Image