🤯 Can explanations of intentionally biased AI models still help clinicians? Data strategies and more

Hiya 👋

We’re back with another edition of the digital pharmacist digest!

Here are this week's links that are worth your time.

Thanks for reading,
Kevin

👓 Currently Reading: Elon Musk by Walter Isaacson

📖 What I'm reading

Technical Debt is not real
👨‍💻 Product management

"In the realm of software development, the inception of a project is often marked by a paradoxical certainty: we know the least about the project at its beginning. This is an inherent trait, not a flaw, of the developmental process.

However, this pair of ideas has a complex knock-on effect. When a new requirement emerges that is antithetical to our previous understanding of the system, we are faced with a dilemma between two choices. The first is to throw up our hands and declare the requirement impossible “Because of the way it is”—this of course is not realistic. The second is to somehow shoehorn this new requirement into the system by means of a nasty hack...

The concept of refactoring sprints, while seemingly a solution, is often misguided. The root problem often lies not in the code itself, but in the processes that built that code. In cases of what might be termed “corporate technical debt,” a continuous, proactive approach to managing technical debt is more effective than sporadic refactoring sprints. This approach requires a rethinking of how we structure our work—every time we get a feature request that doesn’t fit with the system in its current state, we are in fact getting two stories: change the design of the system to be as though it was always meant to have that feature, and make the actual feature."

Data Strategies that Provide Business Value
📈 Data and 👨‍💻 Product management

"A data strategy must focus on building a healthy data ecosystem and an organizational culture that promote the scalability, pliability, and governance of democratized data and allow the organization to take advantage of artificial intelligence and other emerging technologies...

Products over projects: Today, most companies use a data projects approach, where someone with a particular problem asks that data be acquired, cleansed, and prepared to address that one specific use case. In contrast, a data product approach provides a high-quality, ready-to-use data set that can be easily accessed and used by internal customers throughout the organization and is continuously refreshed so it can be applied to many use cases

“One of the core principles of data as a product is that you start from the business and go to the data versus going from the data to the business,”

Measuring the Impact of AI in the Diagnosis of Hospitalized Patients
💊📈 Patient Safety and 🤖 Artificial Intelligence

"Question: How is diagnostic accuracy impacted when clinicians are provided artificial intelligence (AI) models with image-based AI model explanations, and can explanations help clinicians when shown systematically biased AI models?

Clinicians were shown 9 clinical vignettes of patients hospitalized with acute respiratory failure, including their presenting symptoms, physical examination, laboratory results, and chest radiographs. Clinicians were then asked to determine the likelihood of pneumonia, heart failure, or chronic obstructive pulmonary disease as the underlying cause(s) of each patient’s acute respiratory failure. To establish baseline diagnostic accuracy, clinicians were shown 2 vignettes without AI model input. Clinicians were then randomized to see 6 vignettes with AI model input with or without AI model explanations. Among these 6 vignettes, 3 vignettes included standard-model predictions, and 3 vignettes included systematically biased model predictions...

Although standard AI models improve diagnostic accuracy, systematically biased AI models reduced diagnostic accuracy, and commonly used image-based AI model explanations did not mitigate this harmful effect."

Are you enjoying this digest? It would mean a lot if you'd consider forwarding it on to someone that you think would also appreciate it!