This is a response to @frans-3 below who asked why one app can report 90% health, and another 85%, but also in general to anyone using utilities that interpret SMART data, and to be careful what conclusions you draw.
SMART data is just raw data; the industry and drive makers have (purposefully) left open and undocumented, many of the attributes reported, so they are open to interpretation by both drive manufacturers, and by software like Drive X that interpret the data. Seagate may choose to report one attribute differently than how Western Digital chooses too, for example. The SMART specs are often vendor specific, although some over time have come to a general agreement about what it is reporting, but only as it relates to what raw data is reporting, NOT what it means in terms of predicting future health.
Kind of like a car; 4 people may report that they hear a 'funny' noise when the engine runs, they all agree and report there is 'a noise'. They may disagree what it means or predicts, and then the reality might be the engine could fail tomorrow, or it's nothing, and the engine is fine, it's just a noise but does not effect performance.
To make it even more ambiguous, the SMART spec was developed when hard drives were mechanical; it hasn't been updated by the industry since 2011. with SSDs,, even vendors disagree as to what certain attributes mean or are supposed to report when it comes to SSDs, where some of those attributes don't even make sense. For example, a 'bad sector' on a physical mechanical drive has a real physical location, that is fixed (platter 1, side 2, block 30, sector 5, etc). On a SSD, sectors can be relative, the SSD controller reallocates sectors and blocks all the time, a 'sector' number as the operating system thinks of a sector number, doesn't necessarily correlate to the same physical location on a memory chip at any given time, or over time. The concept of 'sectors' and 'blocks' is a hold over from mechanical drives, it is useful to continue to use such terms so that backward compatibility with old BIOS's, drive software and Operating systems is maintained, but 'under the hood', hidden from the operating system, it's not directly identical to how rotational drives managed such things.
I'm not saying these diagnostics are not useful, but you have to take with a grain of salt, especially with SSDs, because it is all up to the drive maker's interpretation of WHAT to report and WHAT is BEING reported, and up to the software vendor of the diagnostic program, to interpret what the data means, and make predictions from it.
Percentage 'health' reports and predictions, are at best generalized and open to interpretation, at worse, potentially misleading. In your example, you have to ask, what does '90%' actually mean, according to DriveDX? Does it mean, 10% of the sectors are bad? Does it mean out of 50 SMART attributes measured, 90% fell in a 'good' range, and 10% fell within a 'bad' range? Is each attribute weighed equally? (obviously some are probably more significant than others, are they given more weight?) Is it a probability, like 90% chance of rain tomorrow? Like 90% chance of failure? When? Tomorrow? A year from now? Over a length of time? What length of time? If it reports 90% today, and 85% a month from now, does it mean you run out and buy a new drive? If someone is using these drives in a mission critical environment (24/7, minimal down time) would he/she take the extreme and say anything less than 100% healthy is unacceptable? In reality, s/he would be replacing drives CONSTANTLY if they took that position, so it begs the question, so if not 100%, then when if at all? Replace at 90%? 85%? 75%?
If your auto repair guy said your car was '90%' healthy, what would that mean? I think you see the point.