Brandon Jackson, MBA

I mentioned in a previous post that successful VBC organizations must strengthen their data pipelines in order to take advantage of analytics that can drive clinical results. To that end below are my 4 pillars of healthcare data pipeline design. Symptoms of a data pipeline with issues include delays in reporting, inconsistent values across your organizational domains, and a lack of trust from clinical users. 1) Raw Record Primacy - Healthcare data is continuously managed, massaged, and warehoused. When incorporating any healthcare data into your analytic structure favor data that is as close to the original source as possible. Trust fields from the billing data over fields from the warehouse (e.g. UB Type of Bill is of more value than an Inpatient Flag or ED Visit ID; MRN will track patient data through an EHR better than a payer member ID). 2) Transparency - Good analysts must understand the data that is being passed to them at the end of the data pipeline. To enable trust and understanding it is critical to provide accurate transparency as to how the data was processed at each step of the journey. Two techniques I have had success with include 'direct documentation' and data lineage. 'Direct documentation' is my term for using the same tables for both data processing and data documentation. There is no need to maintain separate documentation from your code stack as everything can be converted to a table (or file) driven structure. This prevents the inevitable disconnect between your documentation of the code and the actual operation of the code. This further allows for a data lineage engine that can walk specific fields from raw through analytic datamart to accurately explain how the analytic data was created. 3) Conceptual Design - Many healthcare datamarts suffer from concept agglomeration whereby mutiple fields accumulate over time that represent the same underlying concept. This frequently happens during data ingestion as engineers may be unaware of a field that already exists and mistakenly create another to serve the same need. This can happen at any point during data processing. Be ruthless in your conceptual design and create ontologies and hierarchies that organize the data into higher level concepts such that humans can drill down to the specific field they need when doing data mapping. A place for everything and everything in its place will prevent significant downstream confusion. 4) Fail as Fast as Possible - The key to this is to understand what your data processing algorithm 'knows' about the data at each step in the data pipeline. It is impossible to check PMPMs on raw data so don't design your QC process to only fail at the end. Check control totals and field gaps at the start. Once your data mapping is complete you can add field-specific validation. Your data processing algorithm should be learning about the data at each step of the process and the QC should be designed to fail at each step when possible.

39

6 Comments

📝 Here is a Tiny Tableau Tip for Tuesday regarding workflow.   📃 If you are going to make a verbose entry in a tooltip with lots of editing, do it in a word processor and then paste it to your tip.   ❓ Why?   ⚠ If you accidentally hit Esc or the close X while editing a tooltip, Tableau simply discards all your changes.   😲 No prompt. No chance to undo.   👍 Hope that helps someone.   The example text in the simulated loss below is from a Tableau webpage regarding tooltips.   🏗→🧠 Build to Learn!   💭🚶‍♀️🚶‍♂️ Follow for more.   #tableau #data #analytics #VizoftheRay

15

5 Comments

10 lessons from today as a Director of Analytics: 1. Extremely normalized data can be a huge pain to work with. How many look up tables do you need!? 2. When business teams have written off the data team...there's a big problem. The data team needs to take a close look at what went wrong 3. Leading teams brings a whole new level to when you change jobs. You can feel as though you're letting your team down as you transition out. 4. Park far away from your office and walk into work. Your future self will thank you. 5. Get an activity tracker. You don't realize how sedentary data work is until you measure it. 6. Leading is hard and it's easy to second guess what you've done to this point. Stay calm, if you did it all with good intentions and did your best, you did more than most. 7. Be the person Mr. Rogers wanted you to be. 8. Learn Python at some point. 9. SQL is a redneck that flies the "Don't Tread on Me" flag. Yeah, you know I'm right... 10. Care for your people the best you can. They are your responsibility, don't take it lightly. Stay nerdy my friends!

165

31 Comments

Whoever claims building RAG systems is easy has never built one at scale. 🤯 There is a significant gap in the literature and existing frameworks. RAG architectures are designed for POCs, not for hundreds of thousands of documents or supporting tens of thousands of users. Designing these systems requires a cross-functional team with software engineering and data science skills blended with a product mindset. Earlier this summer, I spoke at the Databricks AI Summit, highlighting guiding principles we are taking at CVS Health for building enterprise-grade RAG systems. Every company is unique, and while there is no one-size-fits-all solution, here are three principles that will help you on your journey: 👉 Spend time on discovery—It's tempting to dive in and start building immediately, but you may get lost in the desert without doing your due diligence and understanding what delivers the most ROI. Our team spent approximately a quarter on discovery. 👉 Design for modularity—The LLM space is changing rapidly, and you want to build reusable components that can easily be swapped out or upgraded when something better comes along. 👉 Humans + Machines—Technology alone can not solve the knowledge search problem. Develop processes and governance to help teams curate the data, which will be ingested into your system. These lessons are just the tip of the iceberg. Watch my full Databricks presentation and leave your comments and questions below to learn more tips and tricks. #RAGSystems #EnterpriseAI #DataEngineering #MachineLearning #AIArchitecture #AI #DataScience #LLM #SoftwareEngineering #AIAtScale #KnowledgeSearch #TechLeadership #AIFrameworks #AIInnovation #DatabricksAISummit #ProductMindset #CVSHealth

62

5 Comments

Chances are you didn't invent the database. But there are still so many ways you can use one to automate what you do. Automating makes it simpler to get things done, gives your work more credibility (you took the time to automate it, so you show that your time is valuable), and primes you and prepares you to make incremental improvements. #automation #efficiency #productivity #dataanalytics

5

👉🏼 Does using artificial intelligence take the person out of personal statements? We can't tell 🤓 Jenna N Whitrock 👇🏻 https://lnkd.in/eQ-PpF7U 🔍 Focus on data insights: - 📊 The ability to identify AI-generated personal statements by residency reviewers was only 59% accurate. - 🤔 Only 10% of participants correctly identified all AI-generated statements, indicating a significant challenge in detection. - 🏆 Participants rated AI-generated statements as "best" 60% of the time, suggesting that these statements are perceived favorably despite being machine-generated. 💡 Main outcomes and implications: - 🚫 A majority

1

🎉 And the results are in... 🎉The best type of Analytics Engineering job opportunity is 🥁 Last week, I posted a poll to uncover the most attractive type of Analytics Engineering role. Here's what you all had to say: 1️⃣ The first data hire — capturing the logic around the business and setting up the data foundations and stack: 37% 2️⃣ Joining an established data team with a mature stack — typically growth or replacement hires: 34% 3️⃣ Joining a data team that's new to the modern data stack or Analytics Engineering: 28% Are you surprised by these results? 🤔 I'd love to hear your thoughts in the comments! 💬 #DataAnalytics #AnalyticsEngineering #DataCareers #PollResults #DataTeam #ModernDataStack

6

Do I need Compendium if a data catalog comes "free" (air quotes intentional) with my EMR? It's a legit question, and one that should be asked of anything that comes "free" with your EMR (a data warehouse platform, dashboard tool, etc.). Basically it boils down to whether you want to control your own destiny with a customizable, scalable solution that you can really make your own, or if you're willing to work under a lot of constraints. We wrote a post about this question—check it out and let me know what you think: https://bit.ly/3WC5ByR

8

Try NLP Questions chat with census AI

2

THE HIDDEN OBSTACLES TO SCALING YOUR DIGITAL HEALTH COMPANY: DATA STRATEGY One of the biggest obstacles to fast growth and scalable processes is the mismatch between the data you have and the data you need. If the data needed to run and scale your business is either non-existent or so messy it might as well not exist, you’re in trouble. New digital health companies often focus so intensely on building the "app" to wow users that they delay creating a well-thought-out data strategy. In the chaos of building a company, data gets dumped into spreadsheets and databases without considering the long-term requirements (data and analytics) needed to run and grow the business. My advice: get to know the terms "data ontology" and "data taxonomy" (but don’t drop these terms in an "all hands" meeting). Here’s a simple breakdown in the context of healthcare: Ontology is about defining data properties, and the connections and relationships between data elements (e.g., data formats and how doctors, patients, treatments, and diseases are connected). Taxonomy is about organizing data elements into a structured list or categories (e.g., how diagnosis codes roll up to Hierarchical Condition Categories or HCCs). For more info, check this out: https://lnkd.in/eXH4igk3 Investing upfront in a well-designed data strategy, with input from all stakeholders is crucial. Yes, modifications will be needed over time, but having a solid foundation with data structures, collection processes, naming conventions, and defined relationships will save you so much time in the long run. Pointe Consulting LLC, Luca de Ioanna, Redesign Health, Thrive Capital, General Catalyst, LRVHealth, .406 Ventures #healthcaredata, #ai, #healthcareai, #datagovernance,

4

How much are people spending on platforms that do data lineage tracing? I was able to build a Jupyter Notebook that uses LLM services to read view SQL and trace the lineage back recursively to through logic, CTEs, subqueries, joins, nested views, and ultimately to the data dictionary to build a lineage JSON structure. That JSON then gets converted to a plain English definition of the field. Very useful! #GenAI

40

5 Comments

For current or aspiring independent data consultants, are you struggling to grow your practice? Would $1M+ in revenue in 12 months be compelling? I'm considering putting together a course on how I did just that. DM me if this would be of value.

6

6 Comments

8 lessons from today as a Director of Analytics: 1. Well engineered data brings a deep happiness to the soul of any analytics professional 2. If your first instinct is to assume someone is being malicious or using you, bury it and assume the best until proven otherwise 3. Simple, descriptive stats can bring just as much joy to a stakeholder as a complex, predictive model can, but for much less effort 4. Sometimes a team member just needs to air their grievances, but be prepared to act 5. Everyone has lives outside work, respect that 6. There is no set roadmap to be an analytics professional 7. Business accumen/understanding + analytical skill = deadly combo 8. Social media, including LinkedIn, set you up for a comparison trap. Don't fall for it, run your own race. Stay nerdy my friends!

106

18 Comments