@michalchik if you have symptom(s), then a differential diagnosis is opened: there are several diagnosis possibilities that should be ruled out by tests until a few "survives" and are "rule-in".
General rules of science are applied here; several null hypothesis are tested for each possible diagnosis.
Treatment can or not be given accordingly to final diagnosi(e)s.
if you have a 65 years old woman, but not in a 16 years old woman if they both look healthy and do not complain of symptoms and no family history!
From a doctor's perspective his or her "default" is "there is not cancer" for the youngest one. But that's not true for the older one, based on epidemiological and $ data.I think I am buying it!
The default in all cases in based on evidence, but in thoses case "no symptoms" is not enough evidence to rule out
@michakchik however if no such epidemiological and $ data exists: "there is no symptoms" is enough to say "there is no disease" is the must likely likely hypothesis, so it becomes the "default"
@hedleypanama There is no basis for claiming there is no disease if there is no evidence, beyond the assumption of the null hypothesis. There always can be disease that you did not detect but you do not treat for it. You only ever assume there is disease when there are probabilistic reasons to infer that disease is probable enough to abandon the null hypothesis. That is why you accept the null hypothesis in a 16 year old but don't in a 64 year old.
The null hypothesis is never used to determine whether or not to gather and examine evidence. It used after there have been attempts to gather evidence. IF there is no evidence, non-existence is the working hypothesis. For example, you come across a person with normal test results and no symptoms. Do you start treating him for AIDS because it is conceivable that he might have an undetectable form that might later kill him?
@michalchik whenever you have a scientific inquiry you make a prediction, however BEFORE you do the experiment you made the assumption that the null hypothesis is true!
@hedleypanama I don't think so. If you started with the assumption that the null hypothesis is true, why would you bother gathering data. The assumption that the null is true, is only invoked when evidence has been gathered and there is no evidence to decide in favor of posited hypothesis. The burden of proof is to gather evidence against the null hypothesis.
During screening like colon cancer, the burden of proof is to gather evidence against the alternative hypothesis: there is no symptoms, but there is disease .
For other diseases like brain and lung cancer is like any other case in natural science: if there no symptons the null hypothesis is true by default and no more more test are required: there is no disease or there is disease and uncover diagnosis can be harmful for the patient.
@hedleypanama You need to think about this at a more abstract level. Why do we screen for colon cancer in people but not in airplanes? The reasons is because we have evidence that colon cancer exists in people not planes. If no such thing as colon cancer had ever been found in people would we screen for it? No, no more than we should treat planes for cancer. The precedence for colon cancer exists so we look for it, then if there is no evidence we default to the null hypothesis.
However, whenever you do a "medical inquiry" (a special scientif inquiry) you need clinical evidence (symptoms) in order to do any test, however the screening does not require them.
Every patient is always free of all diseases that NOT fit the WHO qualifications needed for screening. For them null hypothesis is true, not for the other ones.
Treatment can also be used to test hypothesis, but they require an unclear diagnosis.
@hedleypanama Yes, and you were talking about screening tests. But there is evidence for the use of screening tests. Some people have asymptomatic hypertension, your patient is a person, therefore you have evidence that he may have high blood pressure. You do the inquiry, find no evidence. You assume the null. You don't assume that there may be some form of high-BP that you can't detect.
@michalchik In case of AIDS treatment would be conceivable if the person has risk factors, but (a big one) such treatment is expensive and toxic therefore the risks overwhelm the benefits.
If AIDS medicine were non toxic, cheap and demonstrate that prevent disease spreading, then healthy people would take them (regardless the tests). AIDS medicines demonstrate that are able to prevent transmission in special conditions: delivery and accidental puncture only.
@hedleypanama You may make decisions to treat people on scant evidence if the downsides are low, but you never make the decision to treat people when there is no evidence. Diseases exist, some people have them, we know that. Preventative treatment is based on probabilistic knowledge. OTOH we never assume an asymptomatic person has a disease for which we have no evidence of existence. You have never given a person a medicine against "exploding brain syndrome", you assume it does not exist.
@michalchik if you have symptom(s), then a differential diagnosis is opened: there are several diagnosis possibilities that should be ruled out by tests until a few "survives" and are "rule-in".
General rules of science are applied here; several null hypothesis are tested for each possible diagnosis.
Treatment can or not be given accordingly to final diagnosi(e)s.
hedleypanama 1 year ago
@hedleypanama forget about the planes...
if you have a 65 years old woman, but not in a 16 years old woman if they both look healthy and do not complain of symptoms and no family history!
From a doctor's perspective his or her "default" is "there is not cancer" for the youngest one. But that's not true for the older one, based on epidemiological and $ data.I think I am buying it!
The default in all cases in based on evidence, but in thoses case "no symptoms" is not enough evidence to rule out
hedleypanama 1 year ago
@michakchik however if no such epidemiological and $ data exists: "there is no symptoms" is enough to say "there is no disease" is the must likely likely hypothesis, so it becomes the "default"
hedleypanama 1 year ago
@hedleypanama There is no basis for claiming there is no disease if there is no evidence, beyond the assumption of the null hypothesis. There always can be disease that you did not detect but you do not treat for it. You only ever assume there is disease when there are probabilistic reasons to infer that disease is probable enough to abandon the null hypothesis. That is why you accept the null hypothesis in a 16 year old but don't in a 64 year old.
michalchik 1 year ago
@michalchik agree in all your comments... the null hypothesis does not fit IF the prevalence (number of people sick) is high...
hedleypanama 1 year ago
The null hypothesis is never used to determine whether or not to gather and examine evidence. It used after there have been attempts to gather evidence. IF there is no evidence, non-existence is the working hypothesis. For example, you come across a person with normal test results and no symptoms. Do you start treating him for AIDS because it is conceivable that he might have an undetectable form that might later kill him?
michalchik 1 year ago
@michalchik whenever you have a scientific inquiry you make a prediction, however BEFORE you do the experiment you made the assumption that the null hypothesis is true!
hedleypanama 1 year ago
@hedleypanama I don't think so. If you started with the assumption that the null hypothesis is true, why would you bother gathering data. The assumption that the null is true, is only invoked when evidence has been gathered and there is no evidence to decide in favor of posited hypothesis. The burden of proof is to gather evidence against the null hypothesis.
michalchik 1 year ago
@michalchik
During screening like colon cancer, the burden of proof is to gather evidence against the alternative hypothesis: there is no symptoms, but there is disease .
For other diseases like brain and lung cancer is like any other case in natural science: if there no symptons the null hypothesis is true by default and no more more test are required: there is no disease or there is disease and uncover diagnosis can be harmful for the patient.
hedleypanama 1 year ago
@hedleypanama You need to think about this at a more abstract level. Why do we screen for colon cancer in people but not in airplanes? The reasons is because we have evidence that colon cancer exists in people not planes. If no such thing as colon cancer had ever been found in people would we screen for it? No, no more than we should treat planes for cancer. The precedence for colon cancer exists so we look for it, then if there is no evidence we default to the null hypothesis.
michalchik 1 year ago
@michalchik
However, whenever you do a "medical inquiry" (a special scientif inquiry) you need clinical evidence (symptoms) in order to do any test, however the screening does not require them.
Every patient is always free of all diseases that NOT fit the WHO qualifications needed for screening. For them null hypothesis is true, not for the other ones.
Treatment can also be used to test hypothesis, but they require an unclear diagnosis.
hedleypanama 1 year ago
@hedleypanama Yes, and you were talking about screening tests. But there is evidence for the use of screening tests. Some people have asymptomatic hypertension, your patient is a person, therefore you have evidence that he may have high blood pressure. You do the inquiry, find no evidence. You assume the null. You don't assume that there may be some form of high-BP that you can't detect.
michalchik 1 year ago
@michalchik In case of AIDS treatment would be conceivable if the person has risk factors, but (a big one) such treatment is expensive and toxic therefore the risks overwhelm the benefits.
If AIDS medicine were non toxic, cheap and demonstrate that prevent disease spreading, then healthy people would take them (regardless the tests). AIDS medicines demonstrate that are able to prevent transmission in special conditions: delivery and accidental puncture only.
hedleypanama 1 year ago
@hedleypanama You may make decisions to treat people on scant evidence if the downsides are low, but you never make the decision to treat people when there is no evidence. Diseases exist, some people have them, we know that. Preventative treatment is based on probabilistic knowledge. OTOH we never assume an asymptomatic person has a disease for which we have no evidence of existence. You have never given a person a medicine against "exploding brain syndrome", you assume it does not exist.
michalchik 1 year ago