Diagnosis is still left to the subjective assessment of the practitioner before prescribing powerful psychotropics that DO change brain chemistry in unpredictable ways, and cause a host of side effects often worse than the original symptoms.
So the argument today that psychiatry lacks any scientific basis and rigor for its practices is still very valid and legitimate.
If I do a study showing that handwashing in surgeons measurably reduces risk of infection, does the fact that I don't know about microbes make lacking in scientific rigor and remove its scientific basis?
Mind: I'm not arguing that science doesn't make mistakes. Science is a mistake-making-and-correcting process. But there are certain fields of science, somewhat more so in the social than physical and biological sciences, in which a widely-adopted plausible theoretical understanding seems missing.
Science seems to move through stages, of observation and categorisation, to individual models of localised predictive (or descriptive) understanding, to a fundamental theoretical underpinning.
Geology and biology are interesting cases, each being effectively ancient, but for which the establishment of the underlying theory lies largely within human memory.
In the case of biology, taxonomic classification, Darwin's theory of evolution based on variation, inheritence, and selection, Mendelian genetics, and finally the discovery of the structure and reading of DNA and RNA by Watson, Crick, and others (1950s) cast the final links in the chain. We're strengthening those, but the underlying model seems largely complete.
(I'd count concepts of dissipative systems as largely conformant with this model, though important additions.)
For geology, similarly, there was the notion of stratigraphy, which established an ordering but not a specific timescale for geology. Understanding of that didn't occur until the work of Rutherford, Soddy, Holmes, and Boltwood, largely in the first decade of the 20th century.
At the time, the leading scientific value for the age of the Earth was 20 millions of years, by William Thomson, Lord Kelvin (reduced from an earlier estimate of 100 million years). Rutherford and Boltwood's initial measurements based on radioactive decay demonstrated a timescale of billions of years, subsequently refined to the present value of 4.6 billion +/- 1% in the 1950s.
That still didn't explain much geological activity, particularly vulcanology, earthquakes, uplifts, and subsidence (though sedimentation and erosion were well understood). That required the notion of plate tectonics and continental drift, formally adopted only in 1965-1967 (by various conferences / professional bodies).
Plate tectonics, driven by residual thermal heat of formation and radioactive decay in the Earth's core and mantle is now considered the theoretical underpinning of all of geology. It's formal adoption is 55 years old, for a study that's existed since the time of ancient Greece.
The social sciences -- sociology, psychology, political economy, and political science -- lack any such empirically demonstratable and falsifiable theoretcal underpinning. To a large extent they've resisted adopting one.
That last point isn't uncommon -- biology certainaly resisted evolution, and geology plate tectonic (see Naomi Oreske's works on this topic, themselves fascinating studies of the evolution of scientific theory: https://www.worldcat.org/search?qt=worldcat_org_all&q=au%3Ao...).
And as with pre-genetic biology, pre-tectonic geology, pre-Newtonian physics, and pre-Mendeleevian chemistry, there are useful concepts, models, methods, and mathematical relations in these fields. But not a true unifying theoretic basis.