Learning pathology - Revision history

Mikael Häggström: /* Using AI "chatbots" */ Expanded

2026-01-30T19:33:44Z

Using AI "chatbots": Expanded

← Older revision		Revision as of 15:33, 30 January 2026
Line 129:		Line 129:

	==Using AI "chatbots"==		==Using AI "chatbots"==
	For AI "chatbots" (such as ChatGPT, Gemini or Copilot), you can generally trust a simple answer to a simple and presumably common question. More complex answers generally require you to provide more details together with your question. For example, a question of "what is the risk that this patient has..." would require you to provide a patient summary for a better answer, while avoiding identifiable information.<ref group=notes">See [https://www.hhs.gov/hipaa/for-professionals/special-topics/de-identification/index.html#standard HHS de-identification standards] for what counts as identifiable information</ref>		For AI "chatbots" (such as ChatGPT, Gemini or Copilot), you can generally trust a simple answer to a simple and presumably common question. More complex answers generally require you to provide more '''details''' together with your question. For example, a question of "what is the risk that this patient has..." would require you to provide a patient summary for a better answer, while avoiding identifiable information.<ref group="notes">See [https://www.hhs.gov/hipaa/for-professionals/special-topics/de-identification/index.html#standard HHS de-identification standards] for what counts as identifiable information</ref> '''Avoid leading''' prompts (such as "these features indicate invasiveness, right?") but rather keep them neutral and open-minded.

	==Using question banks==		==Using question banks==

Mikael Häggström: /* Using question banks */ Started

2026-01-30T19:30:53Z

Using question banks: Started

← Older revision		Revision as of 15:30, 30 January 2026
Line 127:		Line 127:

	Be tolerant to encountering many '''repetitions''' of the same content in related articles, since for example a [[cervical biopsy]] and a [[cervical cone]] have similar microscopic evaluation.		Be tolerant to encountering many '''repetitions''' of the same content in related articles, since for example a [[cervical biopsy]] and a [[cervical cone]] have similar microscopic evaluation.

			==Using AI "chatbots"==
			For AI "chatbots" (such as ChatGPT, Gemini or Copilot), you can generally trust a simple answer to a simple and presumably common question. More complex answers generally require you to provide more details together with your question. For example, a question of "what is the risk that this patient has..." would require you to provide a patient summary for a better answer, while avoiding identifiable information.<ref group=notes">See [https://www.hhs.gov/hipaa/for-professionals/special-topics/de-identification/index.html#standard HHS de-identification standards] for what counts as identifiable information</ref>

	==Using question banks==		==Using question banks==

Mikael Häggström: /* Using resources */ Flashy

2024-05-22T18:54:28Z

Using resources: Flashy

← Older revision		Revision as of 14:54, 22 May 2024
Line 76:		Line 76:
	*'''[https://www.google.com/ Googling]''' what you are looking for. You may add the word '''pathology''' or equivalent if results are too clinical or layman-oriented. Before even reading the title of results, first look at the URL. If it is from a reputable organization then you may proceed to check if the title is pertinent to what you are looking for, but if it is from an unfamiliar site then you should only proceed if you don't find a better source among your search results. On such unfamiliar sites, you will need to look for additional proof of reliability such as authorship of the content in order to use it as part of any diagnostic decisions.		*'''[https://www.google.com/ Googling]''' what you are looking for. You may add the word '''pathology''' or equivalent if results are too clinical or layman-oriented. Before even reading the title of results, first look at the URL. If it is from a reputable organization then you may proceed to check if the title is pertinent to what you are looking for, but if it is from an unfamiliar site then you should only proceed if you don't find a better source among your search results. On such unfamiliar sites, you will need to look for additional proof of reliability such as authorship of the content in order to use it as part of any diagnostic decisions.
	*'''[https://openai.com/blog/chatgpt/ ChatGPT]''' is useful in answering specific questions, or generating texts with particular content or format. However, don't make significant diagnostic decisions based on its answers without first verifying it with reliable sources, as it might give a false answer despite appearing confident.		*'''[https://openai.com/blog/chatgpt/ ChatGPT]''' is useful in answering specific questions, or generating texts with particular content or format. However, don't make significant diagnostic decisions based on its answers without first verifying it with reliable sources, as it might give a false answer despite appearing confident.
	You can generally rely on online articles for presumably '''well-established''' topics (such as anatomy, and relatively common diseases). However, for potentially controversial and experimental topics, you generally need to ask yourself if the author has a conflict of interest before using its information for important decisions. In such cases, seek highly reliable sources, such as peer reviewed review articles, or professional organizations and societies.		You can generally rely on online articles for presumably '''well-established''' topics (such as anatomy, and relatively common diseases). However, for potentially controversial and experimental topics, you generally need to ask yourself if the author has a conflict of interest before using its information for important decisions. In such cases, seek highly reliable sources, such as peer reviewed review articles, or professional organizations and societies. In contrast, beware of sites with flashy images, as it indicates a financial interest.

	'''[[Ask a colleague]]''' at least whenever your own memory or a resource search is not enough, and there is a significant risk that you may do something irreversible that will negatively affect a patient.		'''[[Ask a colleague]]''' at least whenever your own memory or a resource search is not enough, and there is a significant risk that you may do something irreversible that will negatively affect a patient.

Mikael Häggström: Not sure

2024-05-09T14:38:12Z

Not sure

← Older revision		Revision as of 10:38, 9 May 2024
Line 3:		Line 3:
	\|author2=		\|author2=
	}}		}}
	</noinclude>The goal of this resource is to make a new pathology trainee able to properly handle at least 70% of cases that are expected at an average general pathology department, including the exclusion of the most pertinent differential diagnoses thereof. Its aim is to present you at least '''one way''' of doing things that is at least adequate for a particular situation, ~~so that although~~ there may be various routines at various locations, adherence to these methods should not result in a worse reprimand from a senior than ''"although that was adequate, that is not how we do it here"''. Still '''ask for help''' whenever needed, such as first time you are doing something, or whenever you are not sure about what to do, especially when doing something potentially irreversible.		</noinclude>The goal of this resource is to make a new pathology trainee able to properly handle at least 70% of cases that are expected at an average general pathology department, including the exclusion of the most pertinent differential diagnoses thereof. Its aim is to present you at least '''one way''' of doing things that is at least adequate for a particular situation. Still, there may be various routines at various locations, but adherence to these methods should not result in a worse reprimand from a senior than ''"although that was adequate, that is not how we do it here"''. Still '''ask for help''' whenever needed, such as first time you are doing something, or whenever you are not sure about what to do, especially when doing something potentially irreversible. When confronted with a question, until you are certain what to answer, rather say something in the line of "I'm not sure, but I'll find out and get back to you".

	==What a pathologist needs to memorize==		==What a pathologist needs to memorize==

Mikael Häggström: /* Using resources */ Seek

2024-04-29T17:51:14Z

Using resources: Seek

← Older revision		Revision as of 13:51, 29 April 2024
Line 76:		Line 76:
	*'''[https://www.google.com/ Googling]''' what you are looking for. You may add the word '''pathology''' or equivalent if results are too clinical or layman-oriented. Before even reading the title of results, first look at the URL. If it is from a reputable organization then you may proceed to check if the title is pertinent to what you are looking for, but if it is from an unfamiliar site then you should only proceed if you don't find a better source among your search results. On such unfamiliar sites, you will need to look for additional proof of reliability such as authorship of the content in order to use it as part of any diagnostic decisions.		*'''[https://www.google.com/ Googling]''' what you are looking for. You may add the word '''pathology''' or equivalent if results are too clinical or layman-oriented. Before even reading the title of results, first look at the URL. If it is from a reputable organization then you may proceed to check if the title is pertinent to what you are looking for, but if it is from an unfamiliar site then you should only proceed if you don't find a better source among your search results. On such unfamiliar sites, you will need to look for additional proof of reliability such as authorship of the content in order to use it as part of any diagnostic decisions.
	*'''[https://openai.com/blog/chatgpt/ ChatGPT]''' is useful in answering specific questions, or generating texts with particular content or format. However, don't make significant diagnostic decisions based on its answers without first verifying it with reliable sources, as it might give a false answer despite appearing confident.		*'''[https://openai.com/blog/chatgpt/ ChatGPT]''' is useful in answering specific questions, or generating texts with particular content or format. However, don't make significant diagnostic decisions based on its answers without first verifying it with reliable sources, as it might give a false answer despite appearing confident.
	You can generally rely on ~~books and~~ articles for presumably '''well-established''' topics (such as anatomy, and relatively common diseases). However, for potentially controversial and experimental topics, you generally need to ask yourself if the author has a conflict of interest before using its information for important decisions.		You can generally rely on online articles for presumably '''well-established''' topics (such as anatomy, and relatively common diseases). However, for potentially controversial and experimental topics, you generally need to ask yourself if the author has a conflict of interest before using its information for important decisions. In such cases, seek highly reliable sources, such as peer reviewed review articles, or professional organizations and societies.

	'''[[Ask a colleague]]''' at least whenever your own memory or a resource search is not enough, and there is a significant risk that you may do something irreversible that will negatively affect a patient.		'''[[Ask a colleague]]''' at least whenever your own memory or a resource search is not enough, and there is a significant risk that you may do something irreversible that will negatively affect a patient.

Mikael Häggström: /* Using question banks */ Example

2024-04-24T12:47:06Z

Using question banks: Example

← Older revision		Revision as of 08:47, 24 April 2024
Line 131:		Line 131:
	To test your knowledge on memorization-worthy information, seek questions that present you with a possible real life situation, as well as all pertinent information you can readily look up before needing to answer the question, so to ensure that it tests you on pertinent information as per the items earlier in this chapter. On the other hand, don't waste time on questions whose answer requires memorization of look-up or practically useless information (except if you want to practice looking up the answers online). There are about 5700 questions in PathPrimer, PathDojo, BoardVitals, and the ASCP Question Bank combined,<ref group=note>As of July 2023, PathPrimer has 2270 Multiple Choice Questions (and has the highest quality of its questions despite an overall bad image quality), PathDojo has 1700, BoardVitals has 1050, and the ASCP Question Bank has 675 MCQs. AceMyPath has multiple whole slide images, but is not really a Qbank.</ref> so you can be picky about which questions to take and which ones to skip.		To test your knowledge on memorization-worthy information, seek questions that present you with a possible real life situation, as well as all pertinent information you can readily look up before needing to answer the question, so to ensure that it tests you on pertinent information as per the items earlier in this chapter. On the other hand, don't waste time on questions whose answer requires memorization of look-up or practically useless information (except if you want to practice looking up the answers online). There are about 5700 questions in PathPrimer, PathDojo, BoardVitals, and the ASCP Question Bank combined,<ref group=note>As of July 2023, PathPrimer has 2270 Multiple Choice Questions (and has the highest quality of its questions despite an overall bad image quality), PathDojo has 1700, BoardVitals has 1050, and the ASCP Question Bank has 675 MCQs. AceMyPath has multiple whole slide images, but is not really a Qbank.</ref> so you can be picky about which questions to take and which ones to skip.

	Consider the '''usefulness''' of each question before spending significant time memorizing its answers or explanations, in order to maximize the useful knowledge you will learn overall. Generally, skip questions that test you on look-up information and practically useless information. Whenever you have a question where you can not picture in what real life situation it would be relevant to remember, then you can generally assume that it never will be. For questions that consists of a useful item ~~together~~ with a more useless one, only learn the the useful part and move on. If you can, flag/mark questions on memorization-worthy information that you did not know, and move on fairly quick on those as well, and later revisit those questions specifically if you have the time, as repetition is generally more efficient than lingering.		Consider the '''usefulness''' of each question before spending significant time memorizing its answers or explanations, in order to maximize the useful knowledge you will learn overall. Generally, skip questions that test you on look-up information and practically useless information. Whenever you have a question where you can not picture in what real life situation it would be relevant to remember, then you can generally assume that it never will be. For questions that consists of a useful item combined with a more useless one, only learn the the useful part and move on. The typical example of that combination is a memorization-worthy appearance of a disease, but then the question asks about a rather random fact that you can easily look up if you know the disease name. If you can, flag/mark questions on memorization-worthy information that you did not know, and move on fairly quick on those as well, and later revisit those questions specifically if you have the time, as repetition is generally more efficient than lingering.

	A common sign that a question will deal with look-up or practically useless information is where it asks which of the alternatives is '''false or correct''' for one particular condition, and the alternatives are relatively heterogenous (such as one dealing with diagnosis and another one dealing with prognosis). The high prevalence of such questions is mainly because they are easy and quick create, because the question creator can simply look up a condition, choose some more or less random facts, and then change small details to create alternatives. It requires no real life experience in everyday pathology practice, and learning from such a question is hence generally not useful for that purpose either.<ref group=note>The highest yield questions are generally those with one teaching point per question. In contrast, for a true/false question to be of decent quality, each alternative should be relevant as per criteria above, but it is more difficult to create than low-quality ones. Since low-quality true/false questions are generated at much higher rate, the vast majority of true/false questions are subsequently low-quality.</ref> The thought processes of real life pathology is generally the opposite of this, because you are generally presented with a specific location, gross appearance and microscopic appearance, and you need to practice on finding the most likely causative condition thereof among many differential diagnoses. Even other kinds of memorization-worthy pieces of information generally have a more limited scope among alternatives (like knowing the optimal staining duration in hematoxylin for a frozen section among alternative durations (see [[Emergent pathology]]). Thus, to practice your knowledge in memorization-worthy pathology, it is generally most efficient to skip false-versus-correct questions, in order to spend your limited time on something more worthwhile. Similarly, questions are in general rather low-yield whenever they deviate from the realistic direction.		A common sign that a question will deal with look-up or practically useless information is where it asks which of the alternatives is '''false or correct''' for one particular condition, and the alternatives are relatively heterogenous (such as one dealing with diagnosis and another one dealing with prognosis). The high prevalence of such questions is mainly because they are easy and quick create, because the question creator can simply look up a condition, choose some more or less random facts, and then change small details to create alternatives. It requires no real life experience in everyday pathology practice, and learning from such a question is hence generally not useful for that purpose either.<ref group=note>The highest yield questions are generally those with one teaching point per question. In contrast, for a true/false question to be of decent quality, each alternative should be relevant as per criteria above, but it is more difficult to create than low-quality ones. Since low-quality true/false questions are generated at much higher rate, the vast majority of true/false questions are subsequently low-quality.</ref> The thought processes of real life pathology is generally the opposite of this, because you are generally presented with a specific location, gross appearance and microscopic appearance, and you need to practice on finding the most likely causative condition thereof among many differential diagnoses. Even other kinds of memorization-worthy pieces of information generally have a more limited scope among alternatives (like knowing the optimal staining duration in hematoxylin for a frozen section among alternative durations (see [[Emergent pathology]]). Thus, to practice your knowledge in memorization-worthy pathology, it is generally most efficient to skip false-versus-correct questions, in order to spend your limited time on something more worthwhile. Similarly, questions are in general rather low-yield whenever they deviate from the realistic direction.

Mikael Häggström: /* Learning clinical pathology */ Intro

2024-03-15T20:34:10Z

Learning clinical pathology: Intro

← Older revision		Revision as of 16:34, 15 March 2024
Line 172:		Line 172:
	For '''molecular pathology''', focus on learning the situations in which molecular workup is needed, rather than the specific derangement. The latter can generally easily be looked up when needed, whereas the former can easily be missed.		For '''molecular pathology''', focus on learning the situations in which molecular workup is needed, rather than the specific derangement. The latter can generally easily be looked up when needed, whereas the former can easily be missed.

	{{Learning microbiology}}		'''Microbiology''': {{Learning microbiology}}
	<noinclude>		<noinclude>
	{{General notes}}		{{General notes}}
	{{Bottom}}		{{Bottom}}
	</noinclude>		</noinclude>

Mikael Häggström: /* Learning clinical pathology */ Microbiology

2024-03-15T20:33:18Z

Learning clinical pathology: Microbiology

← Older revision		Revision as of 16:33, 15 March 2024
Line 170:		Line 170:
	<noinclude>Still, if you get the task as a trainee to make a draft of a bone marrow, see '''[[Bone marrow]]'''.</noinclude>		<noinclude>Still, if you get the task as a trainee to make a draft of a bone marrow, see '''[[Bone marrow]]'''.</noinclude>

	For '''molecular pathology''', focus on learning the situations in which molecular workup is needed, rather than the specific derangement. The latter can generally easily be looked up when needed, whereas the former can easily be missed.		For '''molecular pathology''', focus on learning the situations in which molecular workup is needed, rather than the specific derangement. The latter can generally easily be looked up when needed, whereas the former can easily be missed.

			{{Learning microbiology}}
	<noinclude>		<noinclude>
	{{General notes}}		{{General notes}}
	{{Bottom}}		{{Bottom}}
	</noinclude>		</noinclude>

Mikael Häggström: /* Using this resource */ Choose workplace

2024-02-29T22:28:10Z

Using this resource: Choose workplace

← Older revision		Revision as of 18:28, 29 February 2024
Line 124:		Line 124:
	Also, this resource assumes that you use additional sources when needed (such as [https://publications.iarc.fr/Book-And-Report-Series/Who-Classification-Of-Tumours The WHO Classification of Tumors] or others in the ''Using resources'' above). Therefore, it will not simply repeat such information, but may display external links to further information.		Also, this resource assumes that you use additional sources when needed (such as [https://publications.iarc.fr/Book-And-Report-Series/Who-Classification-Of-Tumours The WHO Classification of Tumors] or others in the ''Using resources'' above). Therefore, it will not simply repeat such information, but may display external links to further information.

	This resource also assumes that you will '''ask''' a colleague for advise if you see an unfamiliar presentation, and any literature lookup thereof still does not explain it. Therefore, this resource will focus on the most common presentations and the main differential diagnoses thereof, which may include rare conditions if the clinical management thereof would be significantly different. On the other hand, this resource omits rare conditions that rarely look similar to common presentations. Therefore, keep in mind that there are many such rare conditions that you do not know about, so keep asking colleagues whenever a presentation remains strange to you.		This resource also assumes that you will '''ask''' a colleague for advise if you see an unfamiliar presentation, and any literature lookup thereof still does not explain it. Therefore, this resource will focus on the most common presentations and the main differential diagnoses thereof, which may include rare conditions if the clinical management thereof would be significantly different. On the other hand, this resource omits rare conditions that rarely look similar to common presentations. Therefore, keep in mind that there are many such rare conditions that you do not know about, so keep asking colleagues whenever a presentation remains strange to you, and choose initial workplaces where you have experienced people to ask if needed.

	Be tolerant to encountering many '''repetitions''' of the same content in related articles, since for example a [[cervical biopsy]] and a [[cervical cone]] have similar microscopic evaluation.		Be tolerant to encountering many '''repetitions''' of the same content in related articles, since for example a [[cervical biopsy]] and a [[cervical cone]] have similar microscopic evaluation.

Mikael Häggström: /* What a pathologist needs to memorize */ material

2023-07-13T21:17:51Z

What a pathologist needs to memorize: material

← Older revision		Revision as of 17:17, 13 July 2023
Line 52:		Line 52:
	\| '''Avoiding''', or '''skipping''' to potentially useful parts.		\| '''Avoiding''', or '''skipping''' to potentially useful parts.
	\|}		\|}
	[[File:Cover to cover reading versus problem or project based learning.png\|thumb\|560px\|Example of problem/project based learning versus reading cover to cover. The problem/project based learner may memorize a smaller amount of total information due to spending time searching for the optimal ~~information~~ across various sources, but will likely learn more useful items for real world scenarios, and will likely be better at knowing where to find information when needed.<ref>Image by Mikael Häggström, MD, using source images by various authors. Source for useful context in problem-based learning: {{cite web\|title=Understanding Medical Education - Problem-based learning\|author=Mark A Albanese, Laura C Dast\|date=2013-10-22\|url=https://doi.org/10.1002/9781118472361.ch5\|website=Wiley Online Library}}</ref>]]		[[File:Cover to cover reading versus problem or project based learning.png\|thumb\|560px\|Example of problem/project based learning versus reading cover to cover. The problem/project based learner may memorize a smaller amount of total information due to spending time searching for the optimal material across various sources, but will likely learn more useful items for real world scenarios, and will likely be better at knowing where to find information when needed.<ref>Image by Mikael Häggström, MD, using source images by various authors. Source for useful context in problem-based learning: {{cite web\|title=Understanding Medical Education - Problem-based learning\|author=Mark A Albanese, Laura C Dast\|date=2013-10-22\|url=https://doi.org/10.1002/9781118472361.ch5\|website=Wiley Online Library}}</ref>]]
	Whether any piece of information is useful versus useless depend mainly on:		Whether any piece of information is useful versus useless depend mainly on:
	*The '''likelihood''' of a situation where the information is useful.		*The '''likelihood''' of a situation where the information is useful.