How Hims Uses AI To Benefit Customers, Physicians, and The Business
A deep-dive from Dr. H exploring Hims' proprietary AI system
The use of AI (MedMatch) and how it benefits Hims’ customers, physicians, and the business
Let’s take a look at how Hims’ AI system, called MedMatch, helps customers, physicians, and the business.
Question from Discord member: “So Andrew Dudum is claiming that Hims is a data business and therefore is hiring a CTO to come up with some great AI solution….But if you think about the data points that Hims is getting, it’s very rudimentary data points such as age, sex, ethnicity, allergies, past diagnostics... Is there anything that I'm missing as to properly weighing the true potential of Hims? Do you think the rudimentary data that Hims is currently collecting has a potential to unlock some kind of next-gen AI medicine prescription capabilities?”
What is MedMatch?
The official explanation is: It’s a proprietary AI-powered service developed by Hims & Hers to enhance personalized healthcare delivery. Introduced in November 2023, MedMatch leverages artificial intelligence and machine learning to analyze vast amounts of anonymized patient data, providing healthcare providers with data-driven insights for more effective and personalized treatment.
What are the real-world applications and implications of using AI for Hims?
MedMatch is in the early-game now but is rapidly expanding. The basic information mentioned (demographics, age, race, medical history) is just that: basic. The value will be when they gather more useful information, use AI to analyze the patient's basic info like demographics, but then combine it with other data to then assist in suggesting a better, more optimal treatment to the patient and physician.
Every aspect of the patient’s medical journey will gather data points. From the very first intake questionnaire, through medication and treatment initiation, routine follow-ups, and treatment end points will gather large amounts of information that can be used to help direct future medical care.
KEY POINTS
How patients benefit from AI:
Matching to appropriate medication based on their demographics, previous medication usage history, which medicines would benefit the most based on their symptoms
More accurate treatment / medication recommendations. Less ‘trial and error’. Can save patient money, time, and unnecessary side effects.
Enhanced, personalized treatment plan: ex: custom titration schedules if pt is sensitive to nausea
More efficient and streamlined physician-patient interactions: reduce unnecessary back and forth conversation, less bottlenecks
Higher medication compliance, better outcomes. Higher patient satisfaction.
How physicians benefit from AI:
More personalized treatment recommendations: from making initial treatment suggestions, to custom titration schedules, and side effect management
Improve efficiency of care (less time spent digging for specific information, more time spent with management and overall picture)
Improved treatment outcomes using data driven decision making: avoiding serious side effects, decreasing poor medical outcomes, decreasing litigation risk, increasing physician-patient relationship satisfaction.
How Hims benefits from AI:
Develop new custom medications, compounded medications, or combo medications based on patient trends. Ex: combining statin and ED medication for overall heart health
Increased MOAT in using personalized medications
Increased revenue and profit margins in using custom medications over widely available generic medications
Expansion into new markets and verticals - thyroid, hormonal care
Incorporating new technology - pharmacogenetics testing plus AI for mental health medication management.
A specific AI application involving ED treatment
In the initial intake survey, Hims can collect data like:
Which medicines have you tried before? Sildenafil (generic Viagra), Tadalafil (generic Cialis), Levitra, etc.
How did Sildenafil (generic Viagra) work for you?
How much improvement did you experience with it?
How did the viagra work for you compared to Cialis?
Did it also help you with premature ejaculation?
What side effects did you have with viagra, what side effects did you have with Cialis, how severe were they?
Hims can get really detailed with these questions to build a database of information on how people respond to medicines, and how the medicines compare with each other, and if certain medicines work better for certain demographics over others. Then they can use their compounding pharmacies to make special formulations of the medicine. We've seen them make special versions that combine both viagra and Cialis.
How about if a patient doesn’t know what treatment is best for them? When a new patient comes in and they're not sure which medicine would be best for them, the AI system can review their specific symptoms and severity, match it with a specific Hims-proprietary compounded medicine, and recommend it to the physician.
This helps the patient match to a personalized medication with higher chances of it being effective, rather than the 'old' way of the doctor taking a 'best educated guess' of which medicine may work. It helps the doctor analyze many data points in the history and prescribe the most effective treatment with least potential for side effects.
It takes out the 'trial and error' process which can play out over months of trying one medicine, then another, then another, which leads to low customer satisfaction.
It can guide the patient to Hims' custom formulations which would have a higher margin than basic generic medicines that are available anywhere else. This is already happening with ED meds on Hims. I remember asking patients if there was a specific reason they wanted the custom combination medicine that came through on their order selection. They would respond that it was just what the computer system recommended to them based on their answers on the intake, and it sounded good to them
After reviewing the suggested medication, and reviewing the patient's intake questions, it was indeed a good match. Then I just prescribed. As I got comfortable with the AI system, it made reviewing the patient health information so much easier and efficient. It can mean that it may only take the physician 3 minutes to review, analyze, and come up with a management plan when it may normally take 10 plus minutes to manually do without AI.
An AI application with GLP-1 medications
There’s a lot of debate about when the semaglutide comes off shortage, whether Hims can continue offering personalized semaglutide treatments. Andrew Dudum seems to be confident in doing so.
Here is how AI has helped develop more personalized treatment offerings for semaglutide.
The typical dosing schedule for Ozempic is 0.25mg/week, 0.5mg, 1mg, 2mg.
For Wegovy, it’s 0.25mg, 0.5mg, 1mg, 1.7mg, and 2.4mg.
Per Bayside and this FDA article here:
… there seems to be some interpretation that a 10% variance in medication may allow the compounders to continue compounding even if it is not in shortage.
Hims’ MedMatch / AI system can ask new patients about what their previous experiences with Ozempic or Wegovy were. For example, let’s say they have used Ozempic and made it up to 1mg/week but had issues. They could have found that the 1mg dosage wasn’t strong enough to help with their appetite and weight loss, but when they got bumped up to 2mg, the side effects like nausea and vomiting were TOO strong and prohibited them from continuing treatment.
In this example, Hims can develop custom dosages that are not available on the market: For example they can make dosages closest to 1mg that are 0.8, 0.85, 0.9, 1.13, 1.18 mg (per real world users of Hims’ semaglutide, these are actual dosages available through Hims). Based on the patient’s previous experience, they can be prescribed a 1.13 or 1.18mg dosage, which would be deemed medically necessary, AND just so happens to be outside the 10% variance to appease the FDA.
MedMatch can also inquire about how a patient responds to new medicines. Are they very sensitive to side effects? Does nausea typically bother you a lot when you experience it? This can guide the physician to recommend more customized titration schedules based on this. In the case of a person who is VERY sensitive to side effects, and does horrible with even a small amount of nausea, an extra cautious titration schedule can be used. So rather than going from 0.25mg to 0.5mg to 1mg on Wegovy in the span of 3 months, they can be directed to do 0.28mg, 0.35mg, 0.45mg, for example. This would lead to overall increased medication compliance, less churn, increased safety and patient satisfaction.
An AI application with mental health treatment
You can apply this to any medical treatment. Mental health, depression, and anxiety are notorious conditions that involve a lot of trial and error. If you build up a massive database of next-level information aside from rudimentary demographics, it can take out the guess work for someone if prozac, zoloft, lexapro, wellbutrin, effexor, buspar are going to be the best options. You get the point.
Hims also announced the acquisition of a lab company, Trybe Labs, which could even further open up the application of AI. There is an emerging field called pharmacogenetics testing which looks to analyze a patient’s specific genetic markers which may affect how they absorb, and metabolize medications - this may be important in determining why one medication, say lexapro may not work for them but zoloft does.
The Role of Pharmacogenetics in Personalizing the Antidepressant and Anxiolytic Therapy (LINK: https://pmc.ncbi.nlm.nih.gov/articles/PMC10218654)
When you can combine historical, subjective data (How did medicine X work for you? Did medicine Y work better?) with objective data (lab destiny, genetic marker testing, lab work, demographics like age, sex, etc), you have a vastly large amount of data to help provide optimized medication plans.
How AI helps Hims’ bottom line
Improved physician efficiency and data driven management plans. This would mean physicians can accomplish higher-level tasks, spend less time performing inefficient ones, work to the level of their expertise and lower expenses.
More optimal management plans would mean less patient churn and loss revenue
Identifying new products, treatments, and verticals to grow into.
Increased MOAT in using personalized medications. Over half of Hims’ customers are on some form of custom medication rather than a generic medication. An often cited critique or bear thesis of Hims is that anyone can just start up their own business and sell the medications online. When Hims starts developing more custom and personalized medications based on the data they garner and analzyed by AI, it increases the company’s unique offerings.
A very simple, yet effective usage of this, is the creation of the Heart Healthy medication which combines an ED medication with a statin medicine to protect your heart. Both medications are generic already ( Sildenafil (generic Viagra) or Tadalafil (generic Cialis), plus any statin like atorvastatin (generic Lipitor), but Hims was able to identify a niche where patients who suffer from ED are also at higher risk for heart disease (heart attack and stroke) and test out the effectiveness of taking a combo ED-statin medication a few times a week for better outcomes. The medicines themselves are not unique, but the approach in how they are used is novel and hadn’t been done before.
With an increased library of custom medications, revenue and profit margins can be expected to increase too compared to using widely-available generic medications with little to no profit margin and increased competition.