Hi Hacker News! I’m the founder of DocOne (https://docone.io). DocOne delivers the most relevant physicians for any medical condition based on expertise. We currently cover all US physicians and ~4500 conditions.
Why? The complexity of medicine is accelerating exponentially. Medical knowledge doubles every few months. To keep up, physicians are focusing on narrow groups of conditions. While there are only around 40 traditional medical specialties, in reality there are >500 ultra-specialties.
If there’s one thing we’ve heard over and over from physicians (when they refer their patients) and people (when they search for specialists), it’s that they struggle to identify the right physicians. I’ve personally experienced this. Old-school physician directories (hospital, insurance, 3rd party like ZocDoc and Healthgrades) are essentially phone books. They don’t rely on expertise. They use patient reviews, which have no correlation with expertise. They often sell ads.
That’s why we created DocOne. Our team has a background in data science and medicine. We capture and mine large-scale, up-to-date medical knowledge sources, and use data science (computational linguistics, NLP, shallow and deep learning language models, information retrieval) to build models of medical conditions and to rank physicians according to their relevance to any given condition. Instead of doing the traditional search by physician specialty, people can do a precise search by an individual condition.
We're excited to show you the free and open version of our platform. We've tested it with a narrow group of patients, but we're ready to open it up to a broader user base. To search, start typing a medical condition and select an option from the drop down menu. Search nationwide, or type in a zip code or city/state. Filter by specialty (the platform knows all relevant specialties for each condition, helping the user search for the right type of specialist).
We hope you’ll take a moment to try docone.io and let us know how it goes. Excited to have your feedback! Please leave your thoughts and comments.
The algorithm is flawed. You are prioritizing a number of physicians from corporatized hospital “foundations”. In reality, those often push out private practice physicians with high expertise and hire cheaper newly trained doctors. To raise trust you need to make your rankings a lot more transparent - why be is each doctor ranked so highly? Is it because they have better statistical outcomes (those often don’t take on the hard cases, doctors who save the riskiest patients have worse mortality because they try more)? Is it because they are more published? (A surgeon can have as many papers as you want, but if they don’t do at least 15 whipples per year, they likely deadly for whipple patients (you need to have whipple surgery on your list too). Many medical metrics are counterintuitive. The most important ones are how many of these types of procedures does the Doctor treat per year. And what do other doctors who see that person’s outcomes think of their work. If you want to know which surgeon is good, ask the other surgeons that operate with them.
Agree with most of what you said, except the first sentence. You sound like a physician (and I'm one too, and a data scientist). We take into account all those factors, including relevant procedures/regimens, research activities, quality measures etc. in an extremely granular, disease-specific manner. We don't survey physicians, because that is easily manipulated and it's subjective. Instead, we model nationwide referral network to discover authorities. Thanks for feedback.
Kinda interesting, though maybe you can give some context of who the target market is. Am I as a person with a symptom supposed to decide for myself what type of specialist I need? Or is this for other doctors to find a specialist for their patients?
The problem with #1 of course is "how did I decide what type of specialist I need"? I started by looking for "sleep", not expecting that you had subdivided into different specialists within sleep. Great. Ok, I now have a list of specialists, but how did I know that I needed a specialist in Central Apnea, as an example. If I am qualified to no the specifics of a disease, I'm probably already speaking to a doctor, so ....
Are other doctors the target market? If so, is this the right UI for them? Or would a "LinkedIn for medical" be a better experience, allowing doctors to "connect", review/recommend each other, etc etc
The current UX doesn't actually give any usable info on the doctor, aside from their contact details, on the searches I've looked at. I have no idea what your "Expert Physician", or "Expertise Team" rankings mean. Or the "Clinical Trials" phases. There are no links for me to do anything.
Thanks for great question. Our model is physician/enterprise subscription. This is primarily meant to be sold to health systems and health plans, and used by physicians (who refer and collaborate). However, it is also meant to power patient-facing "doctor search" on the insurance company website or health system website. You as a person would not use this site to guess a disease based on your symptoms. Symptom based search is generally very noisy (there are a few symptoms but thousands of diseases). The uses cases: (a) you know your condition/were diagnosed and looking for the right specialist; (b) there is a suspicion that you have a certain condition. In each case, either your primary physician will refer you or you want to select a specialist yourself (half of all referrals are self-referrals; this is consumerism in healthcare). By the way, try "sleep disorders", the most general term.
You are right about the UI. This site is for beta testing. We are adding About page in a week or so that will explain more what we do and what those scores mean. We haven't invested in enterprise designs yet.
Having done no research on your market, and knowing absolutely nothing, my armchair quaterback with a blown knee and beer gut says...
1) ditch the insurance search thing, unless you can get an insurance company to pay you for it early. It is too difficult for you to describe the benefit here, and is it in the insurance companies best interest? It's barely in the best interest of the patient...because they don't know enough about what they are searching for
2) Have you figured out a moat around the doctors and knowing their expertise? Data is not a moat. Your algorithm is not a moat. You may be on to something, I have no idea. Do doctors use linkedin? if not, is there an opportunity to build a network for doctors. Don't forget, LinkedIn is still backed by search. Should you be building a doctor-to-doctor focused LinkedIn that operates on linking specialties?
Also, this isn't only about doctor to patient, there is also doctor to doctor learning opportunities.
Re clinical trials, a person would learn from their physician whether they qualify and should look for one. It's challenging even for physicians to know which other physicians run trial. Our site indicates for each physician whether or not they participate in trials, and the specific phase.
Okay, so you have Anti-Phospholipid Antibody Syndrome in the system.
How does someone know to search for this, compared to the other, much more common Lupus-related diseases? I mean, even the TV show “House” featured this one.
And how do you not have the internationally renowned Dr. Reichlin in Oklahoma City in the system as one of the world’s leading experts on the disease?
How do you not have Lipodema or Lymphodema in the system? And if you did, how would someone know to search for a doctor with those specialties?
What about fibromyalgia? Or erythromelalgia?
And these are just the rare diseases in my own family and that of my wife’s family.
And finally, how do you determine that someone is an actual expert in a given field, or how they rate in that field compared to others?
Please let me know if you want me to help with searches for your family. As I said, I know this is version 1.0 and not super user friendly yet.
About page goes live in a week and will explain more on expertise scores. Our platforms incorporates the entire medical knowledge in it to compare the expertise that's needed for a certain condition with the expertise that a physician has.
It's not perfect, but I'd love to hear a comparison to other doctor searches. Thanks.
I ran some searches to get a feel for your engine.
It would be good to have conditions grouped by category. For example, a search for lymphoma would ideally be categorized in some hierarchical way.
I would like to know the reasoning behind why physicians were presented. What factors went into choosing them, etc. You need to be able to click on the results and get more info.
I did something similar to this thirty years ago on CDROM. While you have better access to data and can apply new techniques to process the data, this is stil and information retrieval and display problem. The best data is useless if it can't be viewed in a meaningful way.
Agree and thanks for that. We're still thinking how best to map complex disease taxonomies/ontologies into a simple user search and drop-down menu. I like the idea of grouping, it should't be hard.
1. After hitting search there's no indication of activity (while loading. Page remains empty and unresponsive for a while.
2. When updating the search box, [enter] does not work.
3. There is no visual feedback between the list of specialists and the map, so there is no (easy) way to see where a specialist resides.
App:
1. I'd like to see more transparently which specialist maps to what condition (in search results). The "specialty" button seeems a bit non-informative.
2. Why can't I also search for a specialist field directly?
3. Having a DOI included with the medicals specialist's data will make it easier to find more info about the specialist.
All true. This is version 1.0, far from perfect. Good advice, will fix all that soon.
Re the second part: search is by condition, so all specialists ranked relate to that condition. Each condition may be managed by multiple specialties, which is why the Specialties filter is extremely important. For example, you may already have a pulmonologist, but you need an oncologist or a surgeon. You can filter search results to only retain oncologists.
Not sure what you mean "specialist field directly". We will allow search by specialty, meaning that you can select "cardiology" or "oncology". That search won't be condition-specific. It'll rank e.g. oncologists across all cancers.
Like most people, I started with searching my own conditions, but I was surprised that I couldn't find them. Something like Klippel-Feil syndrome/deformity isn't normally rare enough not to make the list. I also couldn't find it by the more generic description "congenital fusion of cervical vertebrae" or variants on those terms.
Having a whole list of conditions starting from the beginning of the alphabet show up as soon as I click the search box isn't terribly useful. Waiting for me to get at least 2 or 3 characters in would be less distracting and more helpful.
Can you explain what the “expertise” and “team” circles are supposed to mean? Also, does light color mean something else than dark color? And what are the “clinical trial” lists for?
It’d also be great if you could offer the link to their website.
Oh, and when I look at the map I see around 10 but when I switch to list, I seem to see all of them (?). This makes me think the map only show the top 10 maybe?
which condition? perhaps they were hidden because only primary specialties are shown when results are first displayed. Click on specialty filter then enable specialties in which your doctors are. however, you are making me think that we should display physicians in all specialties right away when results load, and people can then de-select certain specialties and leave others.
Why? The complexity of medicine is accelerating exponentially. Medical knowledge doubles every few months. To keep up, physicians are focusing on narrow groups of conditions. While there are only around 40 traditional medical specialties, in reality there are >500 ultra-specialties.
If there’s one thing we’ve heard over and over from physicians (when they refer their patients) and people (when they search for specialists), it’s that they struggle to identify the right physicians. I’ve personally experienced this. Old-school physician directories (hospital, insurance, 3rd party like ZocDoc and Healthgrades) are essentially phone books. They don’t rely on expertise. They use patient reviews, which have no correlation with expertise. They often sell ads.
That’s why we created DocOne. Our team has a background in data science and medicine. We capture and mine large-scale, up-to-date medical knowledge sources, and use data science (computational linguistics, NLP, shallow and deep learning language models, information retrieval) to build models of medical conditions and to rank physicians according to their relevance to any given condition. Instead of doing the traditional search by physician specialty, people can do a precise search by an individual condition.
We're excited to show you the free and open version of our platform. We've tested it with a narrow group of patients, but we're ready to open it up to a broader user base. To search, start typing a medical condition and select an option from the drop down menu. Search nationwide, or type in a zip code or city/state. Filter by specialty (the platform knows all relevant specialties for each condition, helping the user search for the right type of specialist).
We hope you’ll take a moment to try docone.io and let us know how it goes. Excited to have your feedback! Please leave your thoughts and comments.