In one of his videos, Dr Mike explains that when diagnosing patients, GPs are on the lookout for two main types of illness:
1) Those which are most common or likely given the symptoms (eg flu is common during the flu season) and
2) Conditions that present high risk to the patient if undiagnosed early even if symptoms are minimal.
For example, many of the symptoms of ovarian cancer are commonly in other, less severe medical problems however the risk of people with ovaries developing it during their lifetime is high, as is the risk of death from it therefore care must be taken to not dismiss symptoms particularly if the patient is over 40/post menopause and thus at higher risk.
Specialised hardware for scanning could help identify cancers without the need to go to a doctor or hospital each time.
A scanner would harnesses supervised image classification (machine learning), through an international public repository of images of benign and malignant tumours/moles/swelling of organs etc as training data for identifying certain cancers. Ideally, as this global repository of training data more training data would make it easier to identify cancers in the early stages, possibly before they would even be detectable to the doctor or patient.
Potentially, people would not even have to make an appointment to go to a doctor or specialist to get a mole checked out - portable scanners could be more easily made available, either for purchase particularly by high-risk individuals or those with a past history or genetic predisposition towards cancers, or available for the general public to use at facilities such as gyms, or non-profit organisations running testing drives to raise awareness, or other institutions seeking to promote health, such as medical aid companies in particular - since oncological treatment is one of the most financially devastating medical costs for patients and thus medical aids often place limits on how much they will cover for that reason and would be strongly incentivised to detect cancers early.