fairMLHealth: Measuring Bias in Healthcare ML is Critical for Fair and Equitable Outcomes

September 02, 2021

KenSci has released an open source tool (fairMLHealth) including tutorials and videos to assist in fair and equitable design and outcomes for healthcare ML

#fairmlhealth #responsibleai

Pillars of Responsible AI in Healthcare

June 23, 2021

How KenSci is meeting the demands of a more trustable and accountable AI in using six distinct pillars of explainability, fairness, robustness, privacy, security and transparency

#responsibleai #pillars

Tutorial: Fairness in Machine Learning for Health Care

August 27, 2020

Responsible ML in healthcare drives adoption of ML and fairness imbedded in healthcare AI/ML tool developments. This tutorial is motivated by the need to comprehensively study fairness in applied ML in healthcare

#fairmlhealth #responsibleai

Award-Winning Clinical Research in Shoulder Outcomes

February 11, 2022

Research from KenSci in partnership with Exactech, won first place for top Orthopaedic implant research at the 2022 Orthopaedic Research Society Annual Meeting

#devicetech #exactech

World's First Machine Based Outcome Score

September 01, 2021

KenSci in collaboration with Exactech developed a new outcome score for assessment of shoulder joint replacement

#devicetech #exactech

Beyond the Usual Suspects: The Role of ML in Suspect Diagnosis

July 15, 2021

Using recommendation system methods, KenSci is reducing the overhead for primary care physicians in identifying the missing diagnosis codes for member care services

#suspectdx #scanhealth

Pillars of Responsible AI

Meeting the demands of more trustable & accountable AI

1-1

Explainability

What factors are driving the model’s prediction?

Suppose a model predicts a patient has a high risk of dying within the next three months but, the physician disagrees with this assessment. In this case, the physician would need to know why the model has predicted to inform action.

2-1

Fairness

Does the model perform similarly on vulnerable group?

What if the model has substantially lower predictive performance for minority or vulnerable patients? Fair ML models are needed to ensure equal treatment of various populations.

3-1

Robustness

Has the model been developed with sufficient data?

If a model is built based on data for population in New York, how will it fair for the African American population in Alabama? Data quality may be different and insufficient data may have been collected. This model would need testing for robustness across cohorts.

4-1

Privacy

Is the model output protected along with other patient data?

AI models can be used to make inferences about a patient which may be harmful if disclosed publicly. Model inversion would allow a malicious entity to infer the values of sensitive attributes, like a rare disease or disability. This may in turn be used to discriminate against the patient by other entities.

5-1

Security

All are data sources monitored and secure?

Data for some models may come from multiple sources. If a malicious person gets access to one of the sources, they can influence how the model gets trained, resulting in incorrect prediction and potential harm. Thus, the security of AI systems in paramount.

6-1

Transparency

Is the AI system and infrastructure transparent?

Suppose a patient is sent to hospice care but the patient survives for more than a year. In this case, the AI system, corresponding pipeline, and infrastructure should be auditable. Accountability ensures that systems are responsible and can be improved in the future.

How it works

FairMLHealth

An open-source tool for fairness assessment of model predictions

Contribute

Watch demo here

Use or download for python

Smart Shoulder Arthroplasty ScoreKenSci in co

KenSci in collaboration with Exactech® introduced a new score using machine learning for assessment of shoulder joint function before the survery.

Read more

A Recommendation System Approach for Suspect Diagnoses

The approach has demonstrated utility for member’s primary care physician with limited overhead when compared to current market alternatives and does not require coordination with outsourced coding experts.

Our Research

Presented at leading forums around the globe