Blissful Life

Category: emr

  • The Ideal Physician AI Assistant

    When I hear “Artificial Intelligence” and “Healthcare” together in a sentence, it is usually never a pleasant thing I’m listening to. There almost always is some kind of reinvention of wheel where Google’s hardware cycles are spent in trying to solve something meaningless.

    For example, it is futile to differentiate between tuberculosis and cancer from an image of the chest where the answer may never lie in the image, but rather in the symptoms of the patient. Even if AI tells the physician that the ECG it is reading is normal (which the physician probably noticed on their own), the physician still has many reasons to refer the patient to a higher center.

    These are isolated examples. But it is the isolation that makes these good examples. AI’s role is not in isolation. AI’s role is in integration. AI (or computers) should come in and fill in where humans struggle – processing large amounts of data. (Processing data, not for the sake of figuring out patterns that humans have easily learned, but for the sake of figuring out patterns, perhaps within an individual, that a human cannot easily learn by going through information)

    AI can be a very good physician assistant. I have previously written about an intelligent EMR. The only barrier to using digital EMRs is the user interface. There are ways to optimize that interface. An intelligent combination of predictive suggestions, tapping rather than typing, reading data from text, etc will help.

    Once physicians can start using EMRs the possibilities are endless. Here is a list of things that come to the top of my mind:

    1. Intelligent to-and-fro symptom/sign/examination suggestion (that physicians can use to not miss important symptoms)
    2. Standard treatment guidelines based suggestions on medications and investigations
    3. Drug interaction checker
    4. Locally relevant and contextual antibiotic resistance patterns
    5. Patient’s past reports based insights, trends, analytics, etc
    6. Medical records exporting, highlighting important information, etc.

    If you are interested in building something like this with me, let me know.

  • Documentation in Medical Records

    I have documented my love of documentation elsewhere. I blog to document my life.

    I’m not perfect at it. Nobody ever can be. Because perfect documentation would take more time than the original act of knowing.

    Imagine. If you were documenting a visit to a nearby tourist attraction. How would you document it perfectly? You could definitely write about it in much detail. But how much detail is enough detail? Would you be writing about everything that you saw on the way? Would you be writing about your thoughts on what you saw? Would you document the planning process? Would you care about other sensations like smell, warmth, etc?

    Recording a video might capture more detail. But a video can’t really capture your thoughts unless you speak into it. Even then it can’t capture your reflections unless you reflect loud while shooting yourself. But how much can you videograph? Where do you store these videos

    Maybe it’s possible to categorize and selectively review any moment from the past using a futuristic memory capture program like shown in Black Mirror. But, seriously, who has the time?

    Perfect documentation is not equal to complete documentation. Documenting all the tiny details would not be relevant at times. But sometimes the tiniest detail can be very relevant.

    This is especially true in medical documentation. Patient’s cousin is a diabetic – relevant. Patient had a day old chicken curry in the morning – relevant. Patient was advised to take so and so medication before food two times a day for 5 days and review if his problems didn’t subside – relevant. Patient is anxious – relevant. Patient reached hospital at 9 am – relevant. Patient teaches in an anganawadi – relevant.

    Documenting all the relevant information is important.

    But, when there is too much information, organization of this information in an accessible manner itself becomes important. Because ultimately, the purpose of documentation is to preserve information for the future so that when one looks back in time, it is possible to accurately interpret history (and avoid controversies. Did Swami Vivekananda’s speech at the Parliament of World Religions get voice recorded?)

    Courtesy: Some CS Professor (Reddit)

    I once saw my consultant Orthopaedician write the timeline of a patient’s visits to hospital and management in her case sheet. This was not really necessary for him to document because none of the information was new or not available elsewhere. Her discharge summaries and OPD case record had all those details. But what the consultant did by summarizing all that in a single page is make it easier for recalling everything at a single glance. The timeline itself added value to the documentation.

    Elsewhere, A Country Doctor writes in his blog:

    Family doctors had the patient’s active problems and their medical,
    surgical, social, family and health maintenance history on the inside
    left of the chart, along with medications and allergies. Our office
    notes, filed in reverse order to the right, were to the point and only
    dealt with the things we had time to talk about that day. But the
    background information was always in view and on our minds. We even used
    to scribble little side notes, like the names of pets and
    grandchildren, hobbies or favorite travel destinations and sports teams.
    The problem list helped us see our patients as individuals, not just
    “the chest pain in room 1”.

    This was an eye-opener for me. I am used to knowing patients’ name by their case record and calling them by their name. But many times than not, I would never know the name of the person accompanying the patient, let alone their children or pets.

    Documentation is an art. It can be perfected only when you know the subject deeply. And when it comes to medicine this amounts to spending quality time with the patient and getting to know them rather than just their illness(es). Like artists, make your documentation picturesque. And people will enjoy it.

  • My Idea of a Perfect Electronic Medical Record System

    The COWs are coming to our hospital.

    No, not these. They’re getting more attention than they deserve.

    Our hospital might soon switch to an Electronic Medical Record system. And this will bring in Computer on Wheels, COW as they’re affectionately called in other hospitals.

    More like this

    While that makes me more happy about where I’m working, it also brings back a lot of ideas I’ve had during medical school. I have seen hard problems for humans that are pretty easy for computers to solve. I have seen processes that could be hastened by leaps and bounds if computers were involved even partially.

    The Perfect Electronic Medical Record System

    The perfect EMR does not just record what the physician or nurse puts in. It is an intelligent assistant that does some thinking of its own and comes back with suggestions and autofills for the physician or assistant.

    For example, when a child comes to you and her mother says she has fever, you start entering “fe…” and the EMR autofills fever. Next you can enter the duration from a dropdown menu. Also associated symptoms can be ticked “yes/no”. As you’re done and move to the next row, the computer automatically populates an entry – “Cough? Yes/No”. If you choose yes, it asks you for characterization.
    If at any point you’re in doubt or do not want to characterize a symptom, you can just delete the autosuggestion and move on to the next line.

    Once you are done with the symptoms and exhausted the negative history that the smart EMR suggested for you, you can enter the examination findings. Again the EMR will suggest for you the most important findings you should not forget to look for based on existing data on what the most common findings are for that particular set of complaints.

    Later, the computer will show you a list of provisional diagnoses based on the data you’ve entered about the patient, and the past set of data the computer has, and even the compendium of knowledge that it potentially has access to. You can reorder the diagnoses if you like.

    Accordingly the computer will suggest investigations and management plans with dosages calculated according to weight or dosage adjustments that are required for special conditions.

    The computer will always suggest and ask you for guidance when in doubt. You can always override the computer, but you can also take computers help in not missing important things.

    Such a system might not be useful for an expert clinician, but it will definitely help a new doctor in emergency rooms late night. Most importantly, the system makes sure that an exhaustive history taking and examination has been done. It also helps in making clinical decisions based on data and evidence.

    Impossible?

    Certainly not. I just discovered that Dr Lawrence Weed, MD has been saying this exact thing since years. He might even have developed such a system already. But EMR systems do not seem to have this kind of intelligence integrated yet.

    The accuracy of simple software powered by big data like Akinator is testimonial to the power of computers when it comes to problems like this.

    Clinical decision making is no holy grail and it will soon be heavily relying on, if not replaced by, artificial intelligence.

    Pictures courtesy pixabay.com