AI in Diagnosis

Artificial Intelligence

Natural Human Intelligence:--

Intelligence is a logical, prior knowledge based application of knowledge and wisdom to tackle, solve or assist in life processes, problem solving and generate further research and evidences. Obviously these are complex processes generated in the neural network and connection of the brain neuro-anatomy. It is this neural communications and emerging actions which we call human intelligence. The fabric in which these actions work in our brain is developed by learning, experiences, anecdotes and our own activity. These are the components of the matrix in which these actions and reactions occur.

The matrix itself is supported by our beliefs—life value systems, theological basis and developed philosophy. The matrix and support are often marred, distorted and modified by personal political agenda, misguided religious teachings and influences of failed systems. This is the downside effects creating inequitable intelligence.

The upside is based on evidence based, researched knowledge and garnished actions based on the correct human (not personal) value system. It can and should be having political leanings not for power but as a means of economic, geopolitical and natural environmentally inclusive activities. It should be with exclusion of personal narrow-minded vision or knee-jerk reactions. This makes it growth oriented. Consequently the human race progresses to a higher level without destroying the surroundings or is “nature” proactive.

Artificial Intelligence:-- It has been human tendency to reduce all life processes into numerical values—the origin of digitalization. At this juncture we as a human race cannot simulate what our brain accomplishes in terms of complexity. The human race has thus split intelligence into smaller and smaller tasks till they can be easily discernable or simplified. Thus:--

1.?????? Knowledge—This is reduced to observed, monitored and measured components. For example-- matter—molecules-atoms-sub-atomic particles-quarks and so on.

2.?????? Education goals-global goals, state goals, institutional, departmental-student goals.

3.?????? Manual dexterity is broken into tasks (components), milestones-benchmarks-competency—skills.

4.?????? Health care—Health Care Delivery—tasks—teachable (teacher skills)- student knowledge gain (student skills)—competency in task completion.

Unfortunately nowhere is the “context” of the concepts evident. The context gives the possibility of using the labelled concepts in real life situations. For example knowledge needs to solve global problems, education for future applications not limited to knowledge gathering per se. Further, , skills need to be developed for the sake of exposing where, how, why and how safely the skills need to be applied and in health care for patient management.

Artificial Intelligence (AI) follows the same philosophy. It is borne out of technology, innovations in task management instead of being “holistic” and end user compliance. For example ChatGPT enhances knowledge, gives it on a silver platter but it is availed without context of their applications albeit with suggestions of their use..

AI is following the same task based approach. So let us see how it is and can be used in health care.

Health Care Philosophy (Allopathy):--

Disease Diagnosis :--It is based on the frequency of occurrence of certain traits in diseased individuals. If a feature is seen in thousands of cases (90% of cases), the feature is termed pathognomonic for the disease. Features occurring in more than 80% of the individuals are cardinal features. Features occurring for more than 60% of the individuals are called indicative features. When combinations of features are seen in an individual it is recognized as “Pattern of the Disease” and is labelled as the diagnosis. This is the nature of diagnostic philosophy. We as Clinicians tend to do this as a mental gymnastics. When individual feature frequencies are positive and combinations of features are causing a conglomerate indication of a disease entity, we call it primary diagnosis. But because these are based on frequencies they are not 100% definitive and thus we also have lesser possibilities—we call them differential diagnosis. The higher the numerical frequency the stronger is the “indicative” diagnosis—a high probability. Artificial Intelligence is based on feeding the computer with such a huge data that the relative error (10% if the frequency is 90%) becomes miniscule as it is a function of the sample size. Effective probability rises and thus AI bases its diagnosis on this statistical high probability. Radio-diagnosis, cancer predictions in AI are a function of the probability statistical estimates.

AI Process:--This is where AI comes in. if you feed data from millions of patients based on ????frequency it gives you the diagnosis.—Radio Diagnosis, Breast Cancer, Other cancers, ????????????predicting risks. All these statistical probabilities are enhanced by the large volume of the data.

The 90% occurrence because of the volume of data, statistically the 10% error gets lessened into miniscule percentages into less than 0.5% ?and thus “voila” AI gives the diagnosis. ?????????????????????????????But humans have variability and thus AI errors are built in these but are too small and ???????????????????????????????????considered to be in acceptable limits !!!

The 95% Confidence Interval and Statistical significance are based on this a”acceptable limits”.

Therapy or Treatment :--The decision making of choice of treatments are also similarly designed. Unfortunately means, equipment, patient choices are doctors skills cannot standardize the treatments hence AI offerings in these matters are minimal.

? Authored by:--?????????????????????????????????????????????????????????????????????

Bharat M. Mody

+91 9724300398

[email protected]

https://ss-bharatmody-ebes.com/blog??????????????

#Diagnosis; # Health Care; #Education; # Artificial Intelligence