Biological age tests based on brain scans for depression are unusable at the individual level (48-62% accuracy). Blood-based phospho-tau217 for Alzheimer's disease does show strong diagnostic performance. 'Biological age test' is not a well-defined single category.
Biological age tests is an umbrella term for measurements that try to determine how 'old' an organ or the brain looks biologically, independent of calendar age. Whether such tests are accurate depends strongly on what they measure and for what purpose.
For detecting major depressive disorder (MDD) via brain scans, the results are frankly disappointing. Researchers trained no fewer than 4 million machine-learning models on three types of brain scans (structural MRI, functional MRI and diffusion tensor imaging). The average accuracy of those models ranged between 48% and 62%. For comparison: flipping a coin gives 50% correct. The distributional overlap between patients with depression and healthy individuals was 87 to 95%, meaning that the brain scans of the two groups are virtually indistinguishable from one another. At the individual level, a biological age test for depression therefore does not yield a usable diagnosis.
This is an important warning signal for the broader market of biological age tests. High technological complexity (millions of models, expensive scans) does not guarantee clinical usefulness. For depression, the evidence is strong that such neuroimaging biomarkers do not work at the individual level.
A very different story applies to the blood test that measures phospho-tau217 as an indicator of Alzheimer's disease. That test shows high diagnostic accuracy across multiple patient groups, both in general practice and in specialised centres. This is a different type of biological marker than a brain age test, but it does fall within the broader category of objective biological measurements for brain disease. The results here are considerably better.
The conclusion is therefore nuanced: 'biological age tests are accurate' is too simple an answer. It depends on the disease, the type of measurement and whether the test has been validated for use at the individual level. For depression via brain scans: no, not usable. For Alzheimer's via blood phospho-tau217: promising and validated in multiple settings. Consumers and healthcare providers would do well to ask, for each test, about the proven accuracy in individual applications, not just about average group results.
Based on two PMID clusters: PMID 38198165 and 35895072 for neuroimaging biomarkers in MDD, and PMID 38422478 for plasma phospho-tau217 in Alzheimer's disease. All claims are associative (no RCTs), but the MDD study is methodologically strong due to the extremely large model space that was examined.