The Effect of Parkinsons on Typing Ability|

Investigating the impact of Parkinson's disease on dexterity and fine motor control through the visualization of typing.

✧

Eric Cheng

Designer

Jerric Jiang

Scatter Plot/Script

Omid Alamdar

Typing Speed/Video

Matthew Budding

Statistics/Aggregate Keystroke

Key Statistics

Brief overview of Parkinson's disease and our dataset.

10 Million people suffer from Parkinsons globally

50% Increase in the US alone, cases per year have gone up from 60,000 to 90,000 (from 2012 to 2022)

1.24 WPM As UPDRS increases typing speed goes down by 1.24 words per minute

60-140 WPM 60% of our Parkinson's cases had a typing speed between 60-140, 52% of our non Parkinson's cases were within the bounds

Typing Speed Comparison

Due to the dibilitating nature of the disease, you might expect a significant difference in typing speed between individuals with and without Parkinson's disease. However, the distribution of typing speeds suggests that the gap is much less pronounced than hypothesized.

Perhaps the Severity of Parkinsons Affects Typing Speed?

The Unified Parkinson's Disease Rating Scale (UPDRS-III) is a clinical assessment of motor impairment in Parkinson's disease. Higher scores indicate greater movement difficulties, including tremors, rigidity, and bradykinesia. This scatterplot explores the relationship between UPDRS-III scores and typing speed. While there is a slight downward trend of -1.24 wpm for every increase of UPDRS, the amount of variability makes it difficult to be certain of a relationship.

"Many people know what Parkinson's looks like… but not the mental anguish it can cause, and the darkest moments we go through."
Guy Deacon, former colonel of the Royal Armoured Corps in the British Army

Keystroke Analysis

Let's investigate the character by character patterns between those with Parkinsons and our control group. In particular, we'll look at their delay between key presses and duration while holding a key. To demonstrate this, we'll showcase one patient with similar typing speed from both of the groups:

pID 1001, who has Parkinson's and a wpm of 118
pID 1002, who does not have Parkinson's and a wpm of 119

Below you can see the typing patterns of these two individuals as well as a graph of the delay and duration corresponding to every key press.

Delay

Flight time from key to key

Duration

Time held down on the key

Parkinson's (pID 1001)

Control (pID 1002)

Key-wise Delay & Duration

Looking at the typing patterns of all participants, we can develop a form of heatmap that shows an aggregated view of the delay and duration for each key. We can calculate the difference between the the Parkinsons and Control group from the heatmap to see if there are any significant differences in typing patterns. You can choose the statistic for which to display the heatmaps below and hover over the keys to see the indicated values.

Choose Statistic:

Below is heatmap of the difference in delay. The darker the red keys mean that patients with Parkinsons take longer to press the key compared to the control group. The darker the blue keys means that patients with Parkinsons take less time to press the key compared to the control group. Note that some values exceed the range of the heatmap.

Delay Differences

Below is heatmap of the difference in duration. The darker the red keys mean that patients with Parkinsons hold the key down longer compared to the control group. The darker the blue keys means that patients with Parkinsons hold the key down for less time compared to the control group. Note that some values exceed the range of the heatmap.

Duration Differences

While there are a few keys in which those with Parkinson's have a lower delay or duration, notably the 'k' key, the majority of the keys show a higher delay and duration when compared to the control group. In other words, those with Parkinson's have greater trouble locating or pressing on just about every key and are also slower to release it.

Aggregated Delay and Duration

To the left is a bar graph displaying the differences in typing behavior between Parkinson’s and non-Parkinson’s individuals by analyzing duration (the time a key is held) and delay time (the time between keystrokes).

Looking at the mean results shows that non-Parkinson’s individuals have a longer hold time, but the median chart suggests a smaller difference, implying variability within the Parkinson’s group. Delay time is consistently longer for Parkinson’s individuals in both charts, indicating slower typing due to motor impairments. The discrepancy between mean and median suggests that outliers in hold time affect the average, highlighting inconsistent typing patterns.

After removing outliers

After removing the outliers using IQR, we notice that both the mean and the median of the duration and delay times are higher for the Parkinson's group. Duration times are heavily reduced from the original mean, resulting in evidence that the Parkinson's group has higher duration times, which is more consistent with our previous findings in the heatmap. Delay times also change but are still have the same relationship as before.

Delay vs. Duration Density Plot

This 2D density plot shows the distribution of delay and duration times for both the Parkinson's and control groups. This reveals that those with parkinsons have more individuals that have higher delay and duration times. However, upon proforming a t-test, we find that the difference in mean delay times is not statistically significant (p=0.43), while the difference in mean duration times is (p=0.02). This suggests that the duration of key presses is a more reliable indicator of Parkinson's disease.

Conclusion

In conclusion, our analysis of the typing data from individuals with and without Parkinson's disease has shown that there are significant differences in typing behavior between the two groups. While the difference in typing speed was less pronounced than expected, the delay and duration of key presses were consistently higher for individuals with Parkinson's disease. Particularly, duration times were found to be significantly higher for the Parkinson's group, suggesting that this metric may be a indicator of the disease. This coincides with the finding in the paper in which we had sourced our data. Our findings highlight the potential of using typing data as a diagnostic tool for Parkinson's disease and provide insights into the motor symptoms of the disease.