That’s because the chart is simply misleading: they collected information on how long the participants slept, and then put the „probability of developing diseases“ next to it.
That’s a fallacy, because most people are in the 7 to 9 hour range, thus the overall probability is lower. The other ranges are usually people with non-nominal lifestyles, which can also very likely impact disease probability.
You can’t just turn around a graph like a mathematical equation... unless it’s sensationalism.