[orc]

"Researchers at Columbia University claim the mChip has a 100 percent detection rate, although there's a four to six percent chance of getting a false positive"

That sounds like a negative result DOES mean you don't have it, but a positive result DOESNT mean you do.

Edit: I guess this is just misleading and you actually can get false negatives.

[alexholehouse]

A negative result DOES mean there are not detectable levels of antibodies. However, while obviously highly linked this is NOT the same as saying you don't have it.

They're correct to say it has a 100% detection rate (assuming it does, but we have no reason to suspect otherwise), but I think this is quite misleading - it's detecting HIV-specific antibodies, not the virus itself.

[orc]

Wow. Yeah, that's pretty misleading. Hope they have a big warning label to go with it.

[wrl]

Better false positives than false negatives. Generally, after testing positive for HIV, there are follow-up tests done to verify the result (usually with a different test from the first) and to check for t-cell count (to determine how damaged the immune system is).

[abeppu]

Does anyone here know anything about what goes into creating a test with a 0% false negative rate? That sounds impressive (even with the noted caveats that it's detecting the antibodies and not the virus).

[stick]

Making a test with 0% false negative rate is a piece of cake:

if (test_taken == true) { test_result = true; }

No negatives means no false negatives. ;)

But seriously, this is the classic trade-off between high sensitivity (no false negatives) and high specificity (essentially no false positives). Usually, the lower you set your threshold for detection the more likely you are to tell some people they have the disease when they actually don't. Where you set that bar depends on what you're trying to accomplish and forces you consider the harms associated with telling people they have disease when they don't or missing disease in someone who does.