Amanda DiTrolio
💡 HTN | Community Brain Trust | 11/14
🧵TOP THREADS OF THE WEEK
In case you missed them, here are highlights of a few interesting conversations from different channels:
Threads included below:
- Best practices for deploying LLMs in a HIPAA-compliant way
- Building a business leveraging patient coaches and navigators
- Choosing between Hubspot and Salesforce for CRM
- Achieving optimal patient matching results with CommonWell and Carequality
- Top claims adjudication software platforms in health plans
1. Best practices for deploying LLMs in a HIPAA-compliant way
Q: What’s the latest thinking on best practices for deploying LLMs in a HIPAA-compliant way? We’re going BAA route
– Zac Schulman | via #buildersask
Thread Summary: HTNers discuss the latest thinking on best practices for deploying Language Model Models (LLMs) in a HIPAA-compliant way. The conversation covers alternative routes, tokenization, and the need for Business Associate Agreements (BAAs) with cloud vendors.
Top Response:
Samir Unni: One “alternative” is to tokenize all identifiers before passing to the LLM. Might call that HIPAA-exempt instead of HIPAA-compliant.
You definitely need a BAA if you use OAI and don’t tokenize. Note that the easier path to use GPT-4 with PHI is likely Azure OAI (BAA with MSFT), rather than OAI proper (BAA with OAI). But it’ll mean you don’t have access to the latest and greatest from OAI, such as what was announced at DevDay (e.g., GPT-4T). However, the reality is that you’ll probably need a BAA with a public cloud vendor anyway - for instance, where are you going to host your VDB? You could of course tokenize before generating embeddings as well - wonder if that’d cause any artifacting in the similarity scores, though.
[Link to Microsoft GPT-4 FAQ] From Nov 7th: Currently, there is no exact ETA we can share when these models will be available in Azure OpenAI.
It’s hard to say without knowing the exact nature of the MSFT-OAI arrangement, but from a strategic standpoint, it’d make sense for OAI to progressively attempt more forcefully to delay/deny access to new products via their not-so-benevolent benefactor in Redmond.
Check out the full HTN Slack convo here!
2. Building a business leveraging patient coaches and navigators
Q: Question based on personal experience - My dad has been a paraplegic most of his life. He’s developed pretty incredible skills at navigating the healthcare system, advocating for his care and preferences and managing his own health. I think he would be an amazing “coach” to others going through similar health complexities (in his case, spinal cord injuries). Has anyone built a business to leverage these types of patient/coaches/champions when it comes to navigating complex health conditions? What are some examples? I’m familiar with forum based solutions (Patients like me) but wondering if there are companies who have taken it a step further to the “coach/navigator” type model.
– Selena Walsh | via #topic-new-company-ideas
Thread Summary: HTNers discuss the concept of leveraging patient coaches and navigators to help others navigate complex health conditions. While there are examples of this concept in condition-specific scenarios like cancer care navigation and maternal care navigation, there haven't been many focused on spinal injuries. The discussion also touches on the challenges of achieving a strong ROI and the role of payers in making it easier. Some members share personal experiences of facilities that provide access to patient coaches and navigators as part of their care.
Top Response:
Tara English: Agree with others - have seen in nonprofit space or as method of hiring navigators / patient advocates in for-profit settings. We have consultants and advisors for our caregiver training work, who developed our training on person-centered caregiving, advocacy, etc. We also partner with Center for Independent Living who have a large focus here.
I think there could be a cool navigator / CHW staffing company play here.
Check out the full HTN Slack convo here!
3. Choosing between Hubspot and Salesforce for CRM
Q: Age old debated: Hubspot vs. Salesforce. Here's some context...
- We are a Series C org that signs <10 deals a year with hospital systems, medical groups, payers, and more.
- The biggest pain points in not having a CRM are manual reporting and difficulty tracking deals.
- Our marketing team uses Hubspot, while our ClinOps team uses Salesforce.
- Salesforce is HIPAA compliant and some of our team believes this is a deal-winner for Salesforce.
- Hubspot has a simpler interface and will cost less for implementation.
Has anyone been in a similar scenario? Any hard-won lessons?
– Anonymous Bot | via #buildersask
Thread Summary: HTNers discuss the pros and cons of using Hubspot and Salesforce as CRM platforms for a Series C organization. The main pain points are manual reporting and difficulty tracking deals. While Hubspot has a simpler interface and lower implementation cost, Salesforce is HIPAA compliant and considered a better enterprise tool. Members share their experiences and offer advice on transitioning between the two platforms and the importance of considering automation, resource management, and HIPAA compliance.
Top Response:
Zach Anderson: I used to work at Salesforce selling into orgs about your size/stage (my answers might be slightly biased based on this exp)
I tend to agree with @Dave Boerner. I actually worked with quite a few orgs that used both Hubspot and salesforce similarly to the mentioned set up.
Why does this post mention not having a CRM when you already have both Hubspot and Salesforce? I'd assume that you should be tracking deals and reporting in one of them?
From my opinion, Salesforce is a more robust system and can handle your current needs and future needs. Reporting and tracking pipeline is table stakes in CRM choices. I believe Salesforce has a better reporting capability to do cross-object reporting.
Three areas to consider are:
1. Are you hoping to automate more workflows or keep things basic? If yes, Salesforce. If no, Hubspot
2. Are you willing to have IT/Marketing Ops/internal resources have to manage multiple systems? If yes, use both as you are. If no, consolidate
3. If you need HIPAA, go with SF
Lastly, feel free to give me a DM and I'm happy to chat privately about your position/choice!
Check out the full HTN Slack convo here!
4. Achieving optimal patient matching results with CommonWell and Carequality
Q: Hi. Does anyone know/can share some information on how to achieve the best possible patient matching results with CommonWell and Carequality in terms of patient addresses? Sometimes I receive typos and invalid addresses from my clients, and in certain cases, our normalization algorithms are changing the data slightly, i.e. the zipcode changes from 12345 to 12346-1234. Or the unit number might disappear - stuff like that. Any help would be appreciated! Also, if you have some useful insight and don't mind helping, please note so in your reply, so I can reach out in private. Thanks!
– Anonymous Bot | via #buildersask
Thread Summary: In this thread, HTNers discuss strategies for achieving the best possible patient matching results with CommonWell and Carequality, specifically in terms of patient addresses. The discussion covers topics such as address normalization, the impact of slight data changes, the use of multiple patient addresses for matching, and the differences in matching algorithms between CommonWell and Carequality. HTNers also explore the role of service adopters/implementers in improving matching, the complexities of the patient matching space, and the limitations of using identifiers like SSN or names+DOB. The thread concludes with a discussion on the potential evolution of the field post-ONC final rule regulations and the possibility of direct integrations with providers for near real-time access to EMR data.
Top Response:
Maylee Wu: Are you connecting with CW and CeQ directly or via a service adopter/implementer like Health Gorilla and Particle? The reason I ask is that some of the service adopters may add additional layer of data normalization and logic in an effort to improve matching.
You're right that CW and CeQ have different matching algorithms, but they also have a different way for how they work. CW uses an MPI/RLS to locate records and uses a combination of deterministic and probabilistic matching logic that uses a combination of computer-assisted matching and human verification. While it does take into account zip code, the weight of zip with and without zip code extension doesn't really have an impact. FWIW, CW has a background process that runs on a periodic basis to normalize addresses to the USPS standard. Also, as long as one of your addresses can match to an existing Person record, then you're likely to get records from the other linked locations.
CeQ does not have an EMPI/RLS so most queries are done via geofencing to an address (whether it's to the patient, provider, or area of choice) or a directed query to a specific organization. Thus, when searching via CeQ, the address is often used for a dual purpose of the region to initially search and then to match to patients within the organizations within the region.
I won't even get into data quality issues or patient disclosure flags that can impact what data you can get back.
Check out the full HTN Slack convo here!
5. Top claims adjudication software platforms in health plans
Q: What are the top claims adjudication software platforms being used at health plans today? Are there larger incumbents, or is it all fragmented/homegrown?
– Anonymous Bot | via #buildersask
Thread Summary: HTNers discuss the top claims adjudication software platforms being used at health plans today. Some of the platforms mentioned include Cognizant Trizetto, Healthedge, Conduent, Gainwell for Medicaid, Mphasis Javelina, S&S, and CapitalRx.
Top Response:
Brendan Keeler: Cognizant Trizetto is one.
Healthedge...Conduent...Gainwell for Medicaid
Check out the full HTN Slack convo here!
🤖HTN KNOWLEDGE BOT
If you have your own question(s) to ask, don’t forget that a good place to start is our HTN Knowledge Bot. It’s our smart search tool that makes it easier to access the wisdom shared within the HTN powered by ChatGPT. You can log in and use it on the website (here) or see how to use it directly in Slack here.
Check out the example ask below!
Member Question:
What are key product components and considerations for a point solution (care delivery via telehealth) to embed in a carrier's plans, specifically commercial (self-funded, level-funded, and fully funded)?
Knowledge Bot Answer:
Based on the conversations and articles provided, I couldn't find any specific information related to the key product components and considerations for a point solution in a carrier's plans. However, I can provide some general considerations for developing a point solution for care delivery via telehealth:
1. Clinical Guidelines and Standards: Ensure that the point solution aligns with accepted clinical guidelines and standards for the specific disease or condition it aims to address. This will help ensure the quality and safety of care delivered through telehealth.
2. Regulatory Compliance: Compliance with relevant healthcare regulations, such as HIPAA, is crucial for protecting patient privacy and maintaining data security. Consider incorporating features and safeguards to ensure compliance with regulatory requirements.
3. Integration with Existing Systems: When embedding a point solution in a carrier's plans, it is important to consider how the solution will integrate with existing systems, such as electronic health records (EHRs) or claims management systems. Seamless integration can help streamline workflows and improve efficiency.
4. Reimbursement and Billing: Understand the reimbursement landscape and ensure that the point solution supports accurate and efficient billing processes. This may involve integrating with billing systems or providing features for capturing and submitting claims.
5. User Experience: Design the point solution with a user-friendly interface and intuitive navigation to ensure a positive user experience for both healthcare providers and patients. Consider features such as appointment scheduling, secure messaging, and virtual visit capabilities.
6. Scalability and Flexibility: Consider the scalability and flexibility of the point solution to accommodate different carrier plans and varying levels of demand. The solution should be able to handle increased usage and adapt to changing requirements.
7. Data Analytics and Reporting: Incorporate data analytics capabilities to track and analyze key metrics related to the point solution's effectiveness and impact on patient outcomes. This can help identify areas for improvement and demonstrate the value of the solution to carriers and employers.
While these considerations provide a general framework, it is important to tailor the point solution to the specific needs and requirements of the carrier and the targeted disease area. Consulting with industry experts and stakeholders can further inform the development of the point solution.
✨HIDDEN GEMS
Here we highlight helpful resources from across the community:
- Provider Data: The Critical Elements to Getting it Right via Joe Mercado – Part 2 of CareJourney's blog series on provide data.