Daniel Cane is co-CEO and cofounder of South Florida-based ModMed®, a healthcare IT firm that’s remodeling healthcare by way of specialty-specific, clever platforms to extend follow effectivity and enhance affected person outcomes.
Based in February 2010, ModMed has grown to over 1,200 workers and has raised over $332 million in whole funding. Recognized for its progressive development as a medical know-how firm, ModMed is often acknowledged each nationally and regionally for its achievements below Daniel’s management. In 2020, the corporate was named one of many Greatest Workplaces within the Nation by Inc. journal. Between 2016 and 2018, the corporate was named one of many fastest-growing firms in North America on the Deloitte Expertise Quick 500™ checklist. Beginning in 2015, the corporate has been named yearly to the unique Inc. 5000 checklist, a prestigious compilation of the fastest-growing non-public firms within the nation.
Are you able to share some insights into your background and the way it has influenced your work at ModMed?
My journey into tech started throughout my undergraduate years at Cornell after I co-founded Blackboard. We remodeled schooling by digitizing class notes and making a platform that gave college students and school unprecedented flexibility and interplay. For me, Blackboard’s success culminated in 2004 with its IPO, and whereas our options had been game-changing in edTech, I couldn’t assist however maintain a watch out for brand spanking new challenges.
One such problem offered itself after I went for a routine checkup with my dermatologist. We had an unimaginable discuss in regards to the struggles of utilizing outdated paper-based methods and methods to repair them. Realizing the bridge between his medical experience and my technical know-how, we determined to group up and create ModMed together with our first digital well being document (EHR) platform.
On the time, some EHRs already existed, however sadly, research typically cited them as one of many main causes of doctor burnout. We took a unique strategy and designed our EHR to adapt the consumer expertise to the particular workflows of a medical specialty. Our flagship cloud-based EHR, EMA, is and continues to be designed by docs, for docs, which has set us aside and defines our secret sauce available in the market. Over time, we’ve expanded our product choices to incorporate a full suite of options that assist medical suppliers simplify and streamline their follow operations and expedite the supply of care.
How do you see the battle for efficient AI in healthcare being received or misplaced with knowledge?
We’re beginning to see an increase within the adoption of AI know-how inside practices to streamline workflows and maximize effectivity. As we transfer into an period of utilizing AI to do extra subtle duties – similar to suggesting remedy or different clinical-support suggestions – it’s paramount to have the fitting knowledge and AI coaching technique in place. AI has the chance to considerably enhance the expertise for sufferers and suppliers and create systemic change that may actually enhance healthcare, however making this a actuality will depend on giant quantities of high-quality knowledge used to coach the fashions.
Why is knowledge so essential for AI improvement within the healthcare trade?
Knowledge is the lifeblood of AI, and poor knowledge high quality will impair an AI’s efficiency, resulting in suboptimal outcomes. This may have dire penalties in a healthcare setting as affected person lives could also be at stake. However a extra doubtless situation is that these adverse experiences may undermine each sufferers’ and suppliers’ belief in AI, slowing down progress and the optimistic impression this revolutionary know-how can have on healthcare.
For instance, within the examination room, AI-enabled ambient listening instruments are designed to counsel content material for scientific notes for the supplier to assessment and approve. Ideally, this could scale back the period of time a supplier spends documenting throughout the EHR and permit for extra high quality time with the affected person. Nonetheless, poor knowledge sourcing and ill-trained AI instruments may have the other impact, leaving suppliers to as an alternative spend an inordinate period of time fixing errors and re-writing notes.
Moreover, bias is a major danger related to AI algorithms, and high quality knowledge can play a key function in mitigating healthcare disparities. AI fashions can study patterns that successfully deal with one affected person inhabitants preferentially in comparison with different populations, together with legally protected teams. By monitoring the info inputs and coaching on sturdy and consultant knowledge, AI outputs will be extra inclusive and correct.
Are you able to elaborate on the sorts of knowledge ModMed makes use of to coach its AI fashions and the way this knowledge is sourced and managed?
At ModMed, we use complete specialty-specific knowledge to assist practice our AI fashions with precision. Over the past 14 years, we’ve created specialty-specific, de-identified structured knowledge units per privateness legal guidelines and are actually leveraging this in-house knowledge to coach our AI fashions. For instance, our ambient listening software ModMed Scribe has been educated for dermatology, our first specialty launch, on thousands and thousands of structured parameters from de-identified affected person information sampled from a set of 500 million affected person encounters.
How does ModMed outline “moral AI” within the context of healthcare?
The potential for AI to have biases or present inaccurate data within the type of “hallucinations” or omissions can impression affected person lives. Because of this, moral AI in healthcare is about setting a excessive customary for accuracy and precision. It means creating algorithms fastidiously and responsibly and utilizing high-quality and numerous knowledge to assist allow extra correct predictions for each consumer.
Moral AI can be about guaranteeing that people stay within the equation. An AI shouldn’t “out physician the physician” however as an alternative scale back the executive burden physicians and their employees expertise to allow them to focus extra on serving to sufferers.
What measures are in place at ModMed to permit AI applied sciences to be developed and deployed ethically?
Our structured knowledge strategy—curating high-quality, consultant coaching knowledge units—helps us make accountable AI a actuality. Related and de-identified knowledge collected from our EHR methods from all kinds of practices gives us with a various set of coaching knowledge that displays totally different affected person populations.
Moreover, our improvement group embraces knowledge cleansing to facilitate accumulating and using high-quality knowledge. This course of permits our groups to determine, rectify, and take away inconsistencies, errors, and lacking values from the info set. By means of this common upkeep, we will persistently replace the AI primarily based on efficiency knowledge, particularly scientific knowledge, the place affected person outcomes will be impacted.
Are you able to talk about the significance of transparency and accountability in AI improvement, particularly in healthcare?
Transparency makes accountability doable, which is why it is such a vital underpinning to any AI resolution in healthcare. Physicians’ high priorities are affected person care and security, so it is no shock that 80% of physicians need to know the traits and options of the design, improvement, and deployment of AI instruments.
Moreover, not all knowledge is created equal. It is necessary to know the place and the way knowledge is saved and sourced and the way often it’s up to date. We’re lucky that since ModMed’s inception, we’ve got been dedicated to a knowledge technique that prioritizes transparency and accuracy. We now have an intensive understanding of our knowledge’s sources and high quality and are assured that our AI integrations will ship appreciable worth to our shoppers.
How is AI being built-in into ModMed’s specialty-specific EHR methods like EMA and gGastro?
Throughout our portfolio, we’ve got been using machine studying for a while and strengthening our funding in superior and generative AI to simplify the enterprise of drugs and expedite high quality care. We’re constructing out a whole AI-powered follow expertise that begins earlier than a affected person walks within the door, extends by way of the examination room, right through to the billing division.
Within the scientific setting, we’re within the ultimate phases of our AI ambient listening pilot program for EMA, which we consider will probably be a game-changer for its downstream performance and advised structured content material. Our AI-powered documentation resolution is designed to streamline the care course of past simply transcription or drafting a SOAP word. Using huge quantities of structured knowledge, we’re coaching our AI fashions to seize important data from doctor-patient conversations and, working alongside our EHR, to counsel related content material for go to notes, together with ICD-10 codes, surgical codes, and prescriptions. This protects physicians valuable time and permits them to spend extra high quality time with their sufferers.
What particular advantages do specialty-specific AI options present to healthcare suppliers and sufferers?
No two medical specialties are alike. They range broadly with the sufferers they see, the situations they deal with, and the medical codes used for reimbursements. AI options have to be tailor-made to accommodate these variations to be efficient in any actually significant means.
For instance, ModMed’s EHRs and AI ambient listening instruments are tailor-made explicitly to every medical specialty, offering extremely related and exact assist to clinicians. Every specialty’s documentation course of requires totally different elements throughout the structured knowledge word, together with distinctive medical codes and terminology. This specialization permits the AI to raised perceive and anticipate the distinctive wants and workflows of various specialty practices, which we consider will end in extra environment friendly implementation, quicker adoption, and better general effectiveness in bettering operational effectivity.
The place do you see essentially the most vital alternatives for AI in healthcare over the subsequent 5 to 10 years?
Sooner or later, AI will undoubtedly permeate almost each facet of healthcare in methods we will’t think about. Already, AI is being harnessed for administrative duties, and within the close to time period, this development will doubtless surge as AI’s worth turns into extra obvious.
I additionally see a future when AI is seamlessly built-in all through doctor-patient interactions, the place the ‘consumer interface’ or UI is just about invisible. As an alternative of right this moment’s screen-based interactions, AI may provide a mix of actuality and augmented actuality. This future state AI may probably analyze well being information to determine essential insights, predicting a affected person’s danger for varied illnesses. The huge quantity of information in medical information presents a possibility for AI to anticipate future care wants and create and assist handle preventive care remedy plans.
This expertise may lengthen past the follow setting and develop into integral to a affected person’s each day life. AI-powered wearables may present personalised assist, reply questions, and schedule appointments amongst different issues. AI may additionally monitor very important indicators remotely, detecting and alerting suppliers to potential well being points. Personalised remedy plans, tailor-made to particular person sufferers primarily based on knowledge and preferences, may develop into the norm.
That is actually an thrilling time for healthcare. The following 5 to 10 years are ripe with alternatives to additional remodel the trade and enhance the affected person expertise.
Thanks for the good interview, readers who want to study extra ought to go to ModMed.