priceMedic

More than 30 million uninsured Americans are being forced to make their healthcare choices without any idea how much their decisions will cost them.  Up until recently, hospitals were not required to publish the prices of their procedures or provide tools for potential patients to estimate their own costs.  On January 1, 2021, a new rule by the Centers for Medicare and Medicaid Services went into effect mandating that every hospital in the United States make all payer negotiated rates available in the form of machine-readable files, but the rule has done little to positively impact consumers.  Prices remain hidden in excel files millions of rows long and massive, non-searchable online databases, rendering consumer-driven price comparison efforts nearly impossible. PriceMedic is building a tool to allow uninsured and underinsured consumers across the country to price shop the medical procedures they need at hospitals close to them. 

Approximately half of uninsured Americans are people of color according to the Brookings Institute.  African and Black Americans were 1.5 times more likely to be uninsured compared to their white/caucasian peers from the period between 2010-2018.  Hispanics were 2.5 times more likely to be uninsured during the same time.  Expansion of Medicaid offers the opportunity to narrow these disparities, but across the board, racial minorities are more likely to fall into the coverage gap.  Rates of coverage gains for these groups consistently lag behind the nationwide average, and coverage losses are felt more heavily. Sixty-two percent of personal bankruptcies filed each year are a result of medical debt.  In 2016, African and Black Americans were 2.6 times more likely to have medical debt.   Insurance disparities accounted for 19.4% of this difference.  Meanwhile, payer negotiated costs for identical procedures can vary as much as 90% at different hospitals within the same region.  PriceMedic is working to reduce the disproportionate burden communities of color face as a result of paying for their healthcare out-of-pocket and the medical debt that frequently follows by allowing all uninsured and underinsured Americans to get their healthcare at the best possible cost.  

PriceMedic is paving the way for real price transparency in the American medical industry, starting with its most vulnerable consumers.  We have compiled a database of the payer negotiated charges for all of the procedures offered at each of more than 500 (and counting) hospitals and outpatient medical facilities across the United States through a combination of data scraping and automated data cleaning techniques.  Currently, Americans who want to know the cost of their healthcare before the book are forced to endure a painful estimate request process through each hospital’s individual billing department.  Most will simply end up booking their procedure without any idea how much the bill will come to. Using priceMedic’s search tool, consumers who will be paying out of pocket can search for the costs of more than 1,000 procedures at hospitals close to them.  Consumers select a location to search from by entering a zip code, city, or address, the distance they are willing to travel from that location, and a keyword or phrase describing the procedure/service they are seeking.  Their search returns a list of hospitals in the area offering that procedure/service ranked by the price they offer it for. 

Our solution is built for every American health consumer who will be paying in whole or part out-of-pocket for their healthcare.  The cost of healthcare in this country is astronomical.  For many members of our target population, the price of care is simply too high.  Three in ten uninsured Americans report going without needed care within the past twelve months.  Individuals without health coverage are less likely to obtain recommended follow-up care for injuries and chronic illnesses.  In many cases, they are making these choices without even being aware of all of their options.  Even with the new transparency rule, price comparison efforts by consumers remain unlikely due to a combination of inaccessible data and medical jargon.  

The lack of price transparency is one of just dozens of ways America’s healthcare system is failing its most vulnerable consumers, and it will take so much more than our single solution to truly make the healthcare they both need and deserve access to them.  We cannot control the price of healthcare, but we do believe that knowledge is power and that because of our solution these individuals will be in a better position to find and select the options that are within reach.      

We are engaging with the populations we eventually hope to serve through a series of private, limited releases designed to inform our iteration efforts leading up to the public launch of our product.  In collaboration with the Kinder Institute for Urban Research at Rice University, priceMedic is working to identify members of our target population residing in Harris County to interview about their own lived experiences interacting with the American healthcare system and test our product at its current stage of development.  Throughout this process, we are fundamentally seeking to understand how price acts as a barrier for them in pursuing the medical care they need or want and how accessible price shopping could potentially alter the course of their decision-making process.  In addition to collecting their unstructured feedback, we will be testing a series of mock scenarios using real hospital pricing data to understand how a more complete understanding of their healthcare options and the costs associated with those options might benefit them.  We will also be collecting their pragmatic feedback on the usability of our product which will be used to help select features for addition and refine the user interface.  

Our solution is built for all under and uninsured Americans who will be charged the payer-negotiated costs made available in compliance with the Centers for Medicare and Medicaid Services 2021 rule for price transparency, but a disproportionately large proportion of this group is made up of people of color.  As such, we expect the implementation of our solution to have a disproportionately positive impact on the healthcare costs of the BIPOC community.  Reducing medical debt and associated bankruptcies will boost access to credit and increase the likelihood of generational wealth transference in communities of color.

At this time, priceMedic is best categorized as a prototype or minimum viable product.  Using our tool at its current stage of development, consumers can compare the prices for individual procedures at more than 500 hospitals/outpatient facilities across the country.  Ultimately, we will be providing both costs for individual procedures as well as cost estimates for commonly bundled procedures and inpatient stays.  For example, with the prototype consumers have the ability to cost compare the price of a C-Section Delivery at different hospitals, but a patient who receives a C-Section will also be billed for their inpatient stay, the cost of sterile materials to perform the C-Section, and transabdominal ultrasounds to prepare for the delivery.  A feature to search for bundled cost estimates will be added before the tool is made available free of cost to the public in a series of state-by-state pilots.  

A stated goal of the price transparency rule is to “begin to narrow price differences for the same services in the same health care markets”.  Through an innovative approach to consumer-driven price comparison, priceMedic may assist in accelerating this transformation. PriceMedic takes a consumer-centric approach to the procedural search process by allowing keyword and descriptive phrase searches.  Most medical price estimation tools require consumers to either identify and search by a specific Current Procedural Terminology (CPT) or Healthcare Common Procedure Coding System (HCPCS) code or only allow for consumers to select from a limited number of preset items.  Open search features like ours are uncommon in the medical billing industry due to the gap between the lay terminology likely used for consumer-driven searches and the medical terminology describing the actual procedures.   To bridge this gap, priceMedic employs a pattern matching algorithm for keywords which is then used to identify the CPT codes of the procedures most likely to be associated with a user’s search, ultimately providing the accuracy of a CPT code search without requiring consumers to deal with the codes directly.  With more than 6,000 hospitals affected by the new price transparency rule and the vast majority of those maintaining their own individual chargemasters, providing the most complete results possible is a challenging task in data aggregation. PriceMedic has developed an API to speed data aggregation and updates by allowing relevant information from individual chargemasters to be dragged and dropped directly into the searchable database of prices. 

Currently, our solution is in the prototype stage and not available for use by the public.  Three close friends and family members have been given access to the prototype to price shop for diagnostic tests and outpatient procedures that they will be paying for out of pocket.  Due to the nature of our solution once made public, it will be accessible to everyone with a wifi connection, but in one year, we expect to have launched with complete searchable price data for every hospital in Texas, Florida, and Arizona providing full features to a total of 8.8 million un/underinsured people living in those states.  In five years, the full features of our solution will be available to our total available market, composed of 30 million uninsured and 28 million underinsured Americans.  Of course, our solution is most useful when consumers need healthcare services beyond what their primary physicians provide.  This will not be the case for everybody every year.  For example, according to the Center for Disease Control, one-fifth of Americans will visit an emergency department and only 7.9% of Americans will stay in a hospital overnight annually.  We estimate that we have the potential to meaningfully impact the healthcare decision-making process for 14% of these individuals each year.  

Once the tool is made live, we will be tracking:

1. The IP addresses of devices used to access our site will allow us to estimate the number of unique and repeat users and how those numbers are growing and changing as we scale.  

2. The location of IP addresses used to make search requests will help us understand the geographic distribution of our users within the United States.

3. A variety of statistics to describe search inquiries performed on our site such as the number of searches performed for individual procedures, the most commonly searched geographic regions, and the radius of interest for specific searches will help us to better understand the healthcare needs of the populations we serve and in doing so will inform iterations of our product and the ongoing addition of features.   

4. A variety of statistics to summarize the results returned for specific searches such as the number of results, average distance from the search location, and Current Procedural Terminology code selected for the keyword search phrase will help us assess how well the medical needs of our target populations are being met in their communities and how well our tool acts to inform users of potential alternatives when those needs are not being met.  

Our team currently consists of 2 founding members who are working full-time.

Within the last ten years, my cofounder and I have lived within 10 minutes' walking distance of three of the largest medical centers in the United States.  When I first learned that my mother needed surgery to remove uterine fibroids, I was so grateful to live across the street from the Texas Medical Center.  My family quickly realized that just because we had access to all the options the medical center had to offer we were not all that much better off.  We spent weeks just trying to determine which hospitals offered the new laser ablation procedure she wanted in place of the hysterectomy her primary care physician had recommended.  Nobody ever offered her a cost estimate, and we didn’t know how to ask for one or even that we could.  My parents were very lucky to have good health insurance, but in the end, the number at the bottom of the bill was essentially a surprise.  Proponents of privatized healthcare in America tout consumer choice as one of the biggest advantages over a socialized system.  But, is it really the consumer’s choice if they are forced to make these decisions without information as critical as how much their healthcare will cost them?  We are excited to use our extensive backgrounds in computer science, statistics, and data science to build towards real price transparency for American healthcare consumers.   

Our founding team was drawn to this competition because of our shared belief that life-changing ideas can come from and be built by anyone, anywhere in the world.  Both founders are biracial minorities and the children of first-generation immigrants, and our project is women-led.  Though it is not often taken into account in assessments of diversity, if we are selected to present in the finals we will likely be the youngest founders by a noticeable margin.  Building towards real price transparency in the healthcare industry is a challenging problem.  We will need the best possible people alongside us as we build, launch, and scale our solution to empower millions of the most vulnerable consumers of American healthcare.  These individuals will come from an incredible variety of situations and backgrounds, so all of our team members will be selected via an open application process rather than closed recruiting or personal connections.  As PriceMedic grows, we will continue to work towards a better future for everyone by avoiding the hiring biases associated with job search platforms, actively minimizing our unconscious biases, and retaining incredible talent with a safe, equitable, and inclusive workplace.  

For unparalleled access to connections and credibility, we would not have otherwise.  We are very young founders.  We are working with less life experience and therefore fewer professional connections we can turn to for advice and assistance.  The medical industry is not well known for entertaining cold calls.  Eventually, introductions to medical industry professionals could help multiply the number of people we are able to serve as well as speed our transition to profitability.  Beyond the connections we make as a result of this competition, being recognized as a Fellow will lend us credibility as we solicit future investments and begin the onboarding process.  Since we are not directly interacting with our target populations like many other teams, measuring our impact will be challenging.  It is easy to measure how many people visit our site or how many searches we support, but it is much more difficult to estimate how those numbers will translate to impact the lives and healthcare choices of those we hope to serve.  In order to truly understand and then magnify our impact, we need metrics, not proxies.  We believe the experts we will meet as a result of this competition will have practical experience to help us ideate and implement more and better solutions for internal impact measurement.  In our cohort of Fellows, we will find a much-appreciated support network and an incredible wealth of knowledge about launching and scaling social ventures.   

One of the largest barriers priceMedic faces is the personal and business relationships that our team needs with hospitals. Without many of these, our job to help hospitals and consumers to our fullest ability is significantly harder. We will need assistance in the public relations sector to form various connections that get us in touch with individuals that know the industry.  Product and service distribution is also critical for priceMedic to get to the people that need it most. We will need to coordinate our social efforts to ensure that our application is reachable and impactful. PriceMedic also makes avid use of AI. This technology has allowed our medical procedure search to be much more user-friendly but it also requires a lot of fine-tuning and effort for the results to be as accurate as possible. Help with sourcing talent and technology will improve our AI-enhanced medical procedure search. This will in turn make our product more accessible to the general public as it will be able to better recognize common colloquial terms and phrases correlated with each procedure.

We are interested in partnering with organizations and individuals that specialize in data aggregation, the policies, legal proceedings, and consumer dynamics of the American healthcare system, and front-end development.  In particular, we believe the impact of our solution would be magnified by a partnership with the MIT Sloan Health Systems Initiative.  With HSI’s assistance, we would like to develop and implement improved analytics to measure how the delivery of healthcare to our target populations is affected by the implementation of our solution and to identify what more we can do to improve their access to healthcare and their health outcomes.  The Data Driven Inference Group or Center for Deployable Machine Learning at MIT CSAIL could assist us as we seek to improve our data aggregation practices.  Finally, a developer who specializes in front-end web and app development could help extend the reach of our product to mobile devices.   

We believe that if we are successful we will transform access to healthcare for millions of Americans.  Healthcare is a human right, and individuals’ ability to make an informed decision about the price of care is a part of that right.  The Centers for Medicare and Medicaid Services’ rule requiring prices to be published was an essential step towards this goal, but the rule alone is and will do very little good if nothing more is done.  The CMS’s rule accomplishes hospital price transparency, but only in the most technical sense of the term.  To price shop, consumers would have to make a list of every hospital in the region, individually source the pricing data for all of those hospitals, and scroll through each datafile to find the cost of their procedure.  It is unreasonable, fraught with potential error, and sets consumers up for failure.  Numerous studies have shown that price transparency can make real change to benefit consumers in the healthcare industry, but only if we find a way to provide consumers with real price transparency.  We are taking a consumer-centric approach to price transparency: no medical jargon, no scrolling through each price entry one by one, all at no cost to consumers. If selected, the funding from the Robert Wood Johnson Foundation Prize will fund the launch of priceMedic in all 50 states, providing price comparison services for every hospital and outpatient facility covered by the Centers for Medicare and Medicaid Services’ rule to nearly 60 million Americans.

When developing PriceMedic, we encountered two critical needs for machine learning and artificial intelligence. The first pertains to the cleaning and predictive completion of the datasets for those that had missing or incorrect values. The second covers the usability of our medical procedure search features.

When creating our centralized database, one of the primary concerns was the aggregation and cleaning of data so that it could be easily input. Since the mandate of the law from the beginning of this year was fairly ambiguous and only required the hospitals to publish their procedures in a “machine-readable” format, we encountered a plethora of different data formats, missing, and incorrect values that would have been required to have grouped the data using traditional algorithms. This led us to the inclusion of artificial intelligence for the cleaning and inference of missing values. With AI we are able to match each hospital's procedure description that was missing a CPT code -- a regulated medical code for procedures that is common across all hospitals -- to one that already had one. This allowed us to produce a completed data set, ready for insertion, by means of predictive imputation. Since around 40% of the hospitals had a complete data set we were able to use these as training data. Additionally, most of the hospital’s procedure descriptions did not stray far from each other thus contributing to a low overall loss for our model.

The second use for AI within our product was for the inference and predictive matching of a user’s input within the procedure search tool. Since most individuals are not knowledgeable on the medical names of procedures, it was critical that our service would be able to interpret their input and accurately match them to suggested results that correlate with it. In addition to that, since procedure names and definitions do not typically include colloquial terms, we were unable to use a keyword search. To solve this, our product uses a live model trained through a user’s search and a proactive follow-up to check if they were content with their results via an easy-to-use widget. The data retrieved after the search is then fed back into the model so that it can easily generate accurate predictions for medical procedures from common affiliated terms and phrases. 

The above two implementations of AI in our product lead us to believe that we are qualified for the AI for Humanity prize. The use of AI-enabled us to aggregate and clean data so that it would be suitable for individuals to accurately compare procedure costs from different Hospitals with a simple search. Additionally, our product will effectively use AI to provide an easy-to-use predictive search tool to enable a large number of individuals to find what they are looking for without having to go through the loopholes of medical definitions. The use of AI provides an accurate, easy-to-use price shopping tool that would not have been accomplished through typical keyword searches and traditional, algorithmic data imputation. 

With the AI for Humanity prize, we will be able to hire talent to help improve our product’s features. It will allow us to decrease loss in the predictive imputation of our data sets and help us improve the friendliness of our search features. Ultimately, the AI for Humanity prize will help priceMedic digitally include millions of uninsured Americans that struggle to find affordable healthcare.