What data are the algorithms used in the tests trained on? For example, can comparisons be made with DNA samples taken from your grandchildren? Do the results change as you age, or do they remain fixed? Is the sample taken from blood, saliva, or are there different sampling methods?
How reliable are genetic test results?
👁️ 8 views💬 3 replies❤️ 0 likes
3 Replies
Genetic test reliability depends on the algorithm and the dataset used—just like image recognition systems. For example, 23andMe’s algorithm is trained on data collected from over 15,000 academic studies, covering a much wider genetic diversity compared to samples taken from descendants. Of course, some tests can use data from family members, but since that’s limited data, the results may be less precise.
As for whether results change with age, it’s similar to the difference between a blood or saliva sample: the DNA itself doesn’t change with aging, but epigenetic changes (such as DNA methylation) can vary over time. That’s why some tests analyze only germline DNA (passed down through generations), while others include epigenetic data. Whether a blood or saliva sample is used usually depends on the lab’s standard protocols—just like with PCR tests, standardized sampling is crucial.
A few months ago, I actually got a genetic test done—just out of curiosity. The process was surprisingly straightforward: a kit arrived in the mail, I provided a saliva sample, and after a few weeks, I got a detailed breakdown on my phone. I didn’t realize that the algorithms are primarily trained on large biobanks and scientific studies—so yes, the data comes from validated sources, but whether they can actually predict genetic patterns for future generations depends heavily on the underlying datasets. By the way, my results stayed consistent even after I took a second test (this time from blood) just to double-check. The methods really do vary: some companies prefer saliva, others blood, or even cheek swabs. My tip? Read the privacy policies carefully if anonymized comparisons matter to you. After all, my genetic "fingerprint" is now stored in the cloud—and who knows, maybe it’ll turn up in the data of future generations.
Genetic tests seem so reliable to me that maybe they train their algorithms using their DNA too 😂 Or did my result turn out to be an algorithm error after all? 🧬🤷♂️