Machine Learning, Ethics, and Change Management

Summary

In this presentation, professors Desi Pachamanova, Vera Tilson, and physician Keeley Matzke share a case study on leveraging machine learning and predictive analytics to enhance operations in a hospital observation unit. Highlighting the challenges of patient triage and the need for effective change management, they detail a real-world scenario where a flawed exclusion list led to increased hospital stays and resource strain. The case covers key themes such as interdisciplinary collaboration, ethical considerations in model variable selection, and the importance of stakeholder engagement. With extensive resources for educators, this multidisciplinary approach aims to improve healthcare efficiency and outcomes.

Takeaways

• 👩‍⚕️ The case focuses on using machine learning and predictive analytics to improve hospital observation unit operations.
• 📚 It offers a multidisciplinary approach, covering predictive analytics, operations management, and change management in healthcare.
• 🏥 The case is based on a real event, emphasizing the importance of appropriate patient transfer from the emergency department to the observation unit.
• 📊 Students can learn from realistic data and scenarios, enabling them to understand practical applications of analytics in healthcare.
• 🔍 The analytics lifecycle framework is utilized, covering stages from discovery to operationalization, allowing for comprehensive teaching opportunities.
• 🧩 The case highlights the importance of stakeholder engagement and change management in implementing analytics solutions in healthcare.
• 📈 By using predictive models, hospitals can reduce waste, improve bed utilization, and enhance patient care quality.
• 💻 Instructors are provided with extensive resources, including data cleaning code and operational models for classroom use.
• 🤖 Ethical considerations in machine learning, particularly regarding variable selection, are emphasized as a key takeaway for students.
• 🗣️ Students express strong interest in discussions around ethical implications and change management strategies related to predictive analytics.

Q & A

What is the main focus of the case study presented?

The case study focuses on optimizing the triage system for an observation unit in a hospital using machine learning, predictive analytics, and change management strategies.

Who are the presenters of the case study?

The presenters are Desi Pachamanova, a Professor of Analytics at Batson College; Vera Tilson, a Professor of Operations Management at the Simon School, University of Rochester; and Keeley Matzke, a physician and the medical director of an observation unit.

What are observation units used for?

Observation units are used for patients who have undergone surgery and need a short stay (24 hours or less) and for patients from the emergency department with less severe illnesses, where physicians need more time to determine if they require inpatient hospitalization.

What challenges did the observation unit face after its opening?

The unit faced issues with a flawed triage system that led to inappropriate utilization of capacity, increased length of stay, and impacted patient flow within the hospital.

How did the presenters plan to address the issues in the triage system?

They proposed creating a predictive decision tool to refine the exclusion list for patient placement in the observation unit, ensuring that the right patients were placed in the right beds.

What framework is used to analyze the triage problem?

The analytics lifecycle framework is used, which consists of six steps: discovery, data preparation, model planning, model building, communication, and operationalization.

What is the significance of stakeholder engagement in this case?

Engaging key stakeholders, such as representatives from various hospital departments, is crucial for successful implementation and decision-making regarding patient placement in the observation unit.

What resources are provided with the case study for instructors?

The case study includes dirty and clean data files, Excel models, instructional materials, and code for data cleaning and decision tree building to aid in teaching.

What ethical considerations did students highlight after working with the case?

Students recognized the importance of ethical considerations in the selection of variables for predictive models, emphasizing responsible use of data and the implications of including social or economic determinants of health.

What key takeaway did the presenters hope to convey about the case study?

The presenters believe the case provides significant educational value through its multidisciplinary approach, real-world context, and comprehensive resources for teaching analytics, operations management, and ethical decision-making in healthcare.