👂 Brain PC Hears You

Good morning! 

Welcome to Healthcare AI News, your weekly dose of the latest developments and headlines in the world of Healthcare AI.

In this issue, we explore:

✅ Headlines: Google unveils MedLM, healthcare-focused generative AI models

✅ Industry: Voice-based AI app reduces time to optimal basal insulin dose in type 2 diabetes

✅ Interesting Reads: What to expect in AI in 2024

✅ Tech: How to stop Dropbox from sharing your personal files with OpenAI

✅ Feature: AI, Health Data, and IP

HEADLINE ROUNDUP

• Google unveils MedLM, a family of healthcare-focused generative AI models (Read More)

• Humana: Class-action suit accuses insurer of using AI to deny care (Read More)

• Can digital twin neighborhoods help tackle health disparities? (Read More)

• Can robots help cure loneliness? (Read More)

• Wolters Kluwer survey: Americans believe GenAI is coming to healthcare but worry about content (Read More)

• Computer made from human brain cells can perform voice recognition (Read More)

TOGETHER WITH ARAGON

Get stunning professional headshots without the hassle! Never pay for a photographer again.

Aragon AI’s tech is built by leading AI researchers and delivers the best results within 90-minutes. Just upload some selfies to see for yourself.

INDUSTRY NEWS

• Voice-based AI app reduces time to optimal basal insulin dose in type 2 diabetes (Read More)

• HHS rule sets AI rules to advance interoperability and transparency (Read More)

• Poison centers see nearly 1,500% increase in calls related to injected weight-loss drugs as people accidentally overdose (Read More)

• Poll shows people seek medical advice from AI and social media instead of their doctors (Read More)

• Hospitals still reluctant to adopt virtual nursing, report says (Read More)

• Patients worry about how doctors may be using AI (Read More)

• Siemens Healthineers introduces Acuson Maple Workhorse ultrasound system (Read More)

INTERESTING READS

• What to expect in AI in 2024 (Read More)

• MIT engineers design a robotic replica of the heart’s right chamber (Read More)

• Apple launches Journal, a new app to reflect on everyday moments and life’s special events (Read More)

TECH NEWS

• Zero downtime Postgres upgrades (Read More)

• Why multi-cloud: the 5 best reasons, according to experts (Read More)

• How to stop Dropbox from sharing your personal files with OpenAI (Read More)

HEALTHCARE AI JOBS

• Senior Machine Learning Engineer

• Multimodal AI Engineer/Scientist

• Senior AI Engineer - Hybrid

THE FEATURE

AI, Health Data, and IP

How AI is complicating the relationship between health data and IP

“My book was stolen.”

This is how social media posts about AI from many outraged authors began recently. They’ve joined the chorus of voices lamenting how they learned LLMs were trained on their work without their expressed permission.

Some big-name authors (think George R.R. Martin and Jodi Picoult) are even suing OpenAI for theft.

It’s pretty clear AI is spurring an important conversation about intellectual property (IP).

But where does health data come into the conversation? 

Recent U.S. copyright hearings have seen calls from artists to require that AI training data be subject to the “three C’s” of IP: consent, credit, and compensation.

These calls already challenge common practice for data extraction around the world. But it gets even more dicey when we consider these same questions for health-related data.

To begin assessing these questions, let’s trace the journey of healthcare data from creation to AI usage and see how IP factors in along the way.

How digital health companies use health data

As people use devices, companies use that data to refine their products and marketing as well as to ideate on new products.

In some cases, companies share this data with third parties.

Especially when it comes to the personal and health-related data of health and wellness apps and devices, this practice can result in controversy. And many questions about who has the right to do what with that data.

Here are a few high-profile cases that illustrate what we mean:

• Flo period tracker charged with lying to users about privacy

• Mental health startup Cerebral shared private health information with advertisers

• A study found third-party tracking on hospital websites

So, is health data IP?

Well, not exactly.

Information related to an individual’s health (like your PHI or smartwatch data) or health-seeking behavior (your web search history) can be unlawful to disclose.

In the US, HIPAA limits what providers and institutions can do with health data without a patient’s consent. And the European GDPR forbids disclosing health data without explicit patient consent under any context outside of direct care. 

However, neither of those conditions qualify it as intellectual property.

What about the right to access your own health data? Does that access indicate the data is your IP? 

We wouldn’t blame you for coming to that reasonable conclusion.

For instance, under HIPAA in the US, individuals have a right to access their PHI under some conditions. With the rise of EHRs and patient-facing online portals, accessing your own health data has never been easier. And importantly, patients have the right to make corrections to the data. The GDPR even permits patients to have their data deleted altogether.

But even under HIPAA, some instances of identifying information do not need to be made available to patients. For instance: customer service records, psychotherapy notes, or information compiled in anticipation of a criminal proceeding.

Overall, the right to access your health data also does not make it intellectual property.

So where and when does health data become IP? We’ll be the first to admit it: The answer is quite complicated. 

So, we’ll present you with two case studies to start making sense of it.

First, let’s look at a household name: 23andMe. The genetic information users shared with the company became their IP. The company then sold the IP for drug development. 

But things look a little different when it comes to clinical visit notes. The physician creates the notes—even if the details are about the patient. They (the clinician) are responsible for the safekeeping of the data, but they don’t own it.

AI training and patient data

And now, it gets more complicated!

When health data is used for training AI models, does that constitute the use of intellectual property? Does that change when the model is generative AI?

Plus, when AI models are trained on health data, what happens to those “three C’s”? Who gets credit for the results? Who needs to consent? And who may be entitled to compensation?

Of course, ideally, identifying information is removed before patient data is used for research. But there have been legal challenges to this assumption, demonstrating how fraught this question can be.

Final thoughts from HAN

Changes in reproductive rights. The rise of telehealth. Pandemic information-sharing norms.

It doesn’t take a hawk to tell we live in an age when digital privacy is continually being redefined.

As we keep scratching our heads and attempting to answer these questions, we’re going to see a battle (or balance) between progress and privacy in healthcare.

We’re interested to see how regulators respond to this challenge. 

In Europe, the proposed European Health Data Space (EHDS) regulation is a welcome step toward better defining data licensing standards and IP definitions within this realm.

We hope, in the process, that patients and consumers get more of a foothold in the new data landscape. At the same time, we believe in the need to encourage the scientific advances the rise of AI promises.

What about you? What questions do you hope to see answered about health data, IP, and AI in the future?

TWEET OF THE WEEK

🌟 Advertise With Us 🌟

Boost your brand amongst Healthcare's influential circle! Our diverse subscriber base boasts top executives, key decision makers, and visionary professionals from leading organizations – the ultimate platform for your brand's success. 🔥

Get in touch today