How would you even know if it is consistently inaccurate? The variability in biological signals might be so high day to day that it could be highly correlated but with frequent pathological cases that make the data useless.