Test-retest reliability of plasma biomarkers and digital cognitive metrics at clinically meaningful timescales
This study evaluated the test-retest reliability of plasma biomarkers and digital cognitive metrics over 0, 3, and 6 months in cognitively unimpaired and early Alzheimer's disease individuals, finding that digital measures demonstrated high reliability without needing repetition, whereas pTau181 showed the poorest reliability and benefited from repeated sampling for improved diagnostic accuracy.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
Imagine you are trying to tune a radio to find a clear station. Sometimes the signal is crystal clear, but other times it's filled with static, making it hard to tell if you're hearing a song or just noise. In the world of medicine, especially for Alzheimer's disease, doctors and scientists are constantly looking for "signals" that tell them if a person's brain is healthy or if it's starting to show signs of trouble. Two of the most promising tools they are using right now are like a blood test and a digital video game. The blood test looks for tiny protein fragments floating in the plasma, while the digital game asks the player to solve puzzles on a screen to see how fast and accurately their brain is working.
The big question scientists are asking is: "How much of the noise in these tests is just random static, and how much is a real change in the person's health?" If a blood test result jumps up and down wildly from day to day without the person actually getting worse, it's hard to know if a new medicine is working. Similarly, if a digital game score changes just because the player had a bad night's sleep or a different mouse, it's not a reliable way to track the disease. To solve this, researchers need to know how "reliable" these tools are when you use them again and again over a few months. They want to know if taking the test twice helps clear up the static, or if the first try was already perfect.
This study, led by a team at the University of Oxford and other institutions, decided to play a game of "spot the difference" with 92 people—55 healthy volunteers and 37 people with early Alzheimer's disease. Over the course of six months, these participants visited the lab three times (at the start, three months later, and six months later). At each visit, they gave a blood sample and played a series of digital cognitive games on the "OCTAL" platform. The researchers then compared the results from visit to visit to see how much the numbers bounced around. They were looking for the "gold standard" tools: the ones that stayed steady and gave the same answer every time, versus the ones that were jittery and unreliable.
Here is what they found, and it's a bit of a mixed bag. When it came to the blood tests, not all the proteins were created equal. One protein, called pTau181, was the most jittery of the bunch. It was like a radio station that kept switching frequencies; its results varied so much that if a doctor only took one blood sample, they might get a misleading picture. In fact, for this specific protein, taking a second blood sample actually helped clear up the picture and made the diagnosis more accurate. However, another protein, pTau217, was much more stable, acting like a strong, clear signal. Interestingly, the ratio of two amyloid proteins (Aβ42/40) was very steady in terms of how much it varied from person to person, but when looking at how it changed within the same person over time, it wasn't as helpful for telling healthy people apart from those with Alzheimer's as the pTau217 test was.
The digital cognitive tests told a slightly different story. These were generally much more reliable than the blood tests. The researchers found that certain tasks, like a visual memory game called the "Oxford Memory Test" (where you have to remember where objects were placed) and a drawing task called the "Rey-Osterrieth Complex Figure" (where you copy a complex shape), were very consistent. If a person played these games three times over six months, their scores didn't bounce around wildly. This suggests that these digital games are excellent tools for tracking changes over time without needing to repeat them just to get a "better" average. Unlike the pTau181 blood test, repeating the digital games didn't significantly improve the ability to tell if someone had Alzheimer's or not; the first try was usually just as good as the second.
The most important takeaway from this research is that we can't treat all these tests the same way. Some, like the pTau181 blood test, are so noisy that you might need to take the test twice to be sure of the result. Others, like the pTau217 blood test and the digital memory games, are so reliable that a single measurement is likely enough to give a clear answer. This is a crucial piece of the puzzle for the future. If doctors know which tools are "steady" and which are "jittery," they can design better clinical trials and make more confident decisions about patient care. They won't waste time repeating tests that are already perfect, and they won't miss important changes because they only took a noisy measurement once. The study didn't prove that these tests are perfect for everyone, but it did map out exactly which ones are ready to be trusted and which ones need a little extra help to be reliable.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.