Volv Global SA

It’s been a pleasure to talk with Jessica Lynn from Patient Worthy about the challenges in rare disease diagnosing and how new technologies can help in this. With Volv Global SA #inTrigue #inClude #driveforchange #rarediseaseresearch #iainhealthare

AATD Conclusions What can be learned from this Machine Learning approach to patient cohort finding? We found: 1/ The Asian, Black or African American, and Hispanic or Latino populations were underrepresented in the AATD-diagnosed patient pool when compared to actual US demographics. 2/ Although racial and ethnic biases were evident, the prediction model accurately identified patients with likely undiagnosed AATD earlier in their diagnostic journey in these subpopulations. 3/ Among patients diagnosed with AATD, there were fewer treated than tested individuals. 4/ Treatment rates were lower in non-White and Hispanic patients. 5 / Evaluating by race and ethnicity, a larger proportion of non-White and Hispanic or Latino patients with likely undiagnosed AATD had asthma. If you would like to see how this process can help your efforts to reduce health disparities in underserved communities, please message us directly.

AATD Study Results - 3 of 3: Evaluating by race (below) and ethnicity (not displayed), a larger proportion of patients with likely undiagnosed AATD had asthma.

AATD Study Results - 2 of 3: The Treatment Disparity - by race and ethnicity among patients with AATD. Among patients diagnosed with AATD: Across all race and ethnicity groups, there were numerically fewer treated than tested individuals (Figures A and B). Treatment rates were lower among non-White patients (Black or African American = 24.53%, other non-White = 21.99%, Asian or Pacific Islander = 14.89%, Figure A) and Hispanic or Latino patients (16.69%, Figure B) compared with White patients (37.86%, Figure A) and non-Hispanic or Latino patients (34.24%, Figure B).

AATD Study Results 1 of 3: Cohort Evaluation ― Potential Diagnostic Disparity. Relative to the US population distribution, non-White and Hispanic patients with AATD (positive cohort), appear to be underrepresented. The machine-learning prediction model may more accurately reflect respective populations of underrecognized AATD. Implicit biases surreptitiously influence judgment and can, without intent, contribute to discriminatory behavior. [Source: Greenwald AG, Dasgupta N, Dovidio JF, Kang J, Moss-Racusin CA, Teachman BA. Implicit bias remedies: treating discriminatory bias as a public health problem. Psychol Sci Public Interest 2022;23:7-40.] Our work suggests that ML based diagnostic support tools may help negate some of these implicit biases. If you would like a copy of the full study in PDF format, please comment below, or message us direct.

Overcoming Racial and Ethnic Biases in the Diagnosis of Patients With Alpha-1 Antitrypsin Deficiency in the United States Using a Machine-Learning Model. The Objective: To develop a prediction model to identify symptomatic patients of different races and ethnicities with likely risk of AATD using claims data from a large US database. Implicit and explicit biases are among many factors that contribute to disparities in health and health care. (Tackling Implicit Bias in Health Care Published July 9, 2022 N Engl J Med 2022;387:105-107 DOI: 10.1056/NEJMp2201180 ) In partnership with Takeda, we took to designing a machine learning process to find likely candidates and overcome racial biases in the detection of this disease that may result in serious lung or liver disease. AATD is largely underdiagnosed, with an estimated prevalence of 100,000 individuals with AATD in the US; however, fewer than 10,000 individuals are diagnosed (Ashenhurst JR, et al. Chest. 2022;161(2):373-381.) Previously, AATD was thought to affect only White individuals of European descent. Recent studies have shown that people of different races and ethnicities have genotypes consistent with those with moderate-to-severe AATD-related lung disease. (Quinn M, et al. Ther Clin Risk Manag. 2020;16:1243-1255. de Serres FJ, Blanco I. Ther Adv Respir Dis. 2012;6(5):277-295.) The Process: Data from the Komodo Health US claims database (April 26, 2016 to January 31, 2023) were divided into “positive,” “negative,” and “target” cohorts. A machine-learning model for detecting AATD was trained on positive and negative cohorts without using codes revealing AATD diagnosis and treatment.The learned model was applied to the target cohort to flag patients with likely undiagnosed AATD. Results: This approach produced a highly performant prediction model capable of detecting undiagnosed people living with AATD, validated by expert clinicians. (For a deeper look at how this unique ML process could be applied to other indications, please message us directly.)

See you all there in Barcelona! Volv are really looking forward to this years event. #WODC2024 #Volv

PHARMACEUTICAL CONTRACT MANAGEMENT LIMITED #PCMG2024 Learn more about the practical reality of AI and and its use in the pharma industry. Together with Tim Hardman Niche Science & Technology Ltd, Benoit Cuyvers ArtiQ. #inTrigue #inClude #driveforchange #innovationinhealthcare Volv Global SA

Health Equity & AI: helping address the issue of bias in healthcare Overcoming Racial and Ethnic Biases in the Diagnosis of Patients With Alpha-1 Antitrypsin Deficiency in the United States Using a Machine-Learning Model • Previously, AATD was thought to affect only White individuals of European descent[2,3] • Recent studies have shown that people of different races and ethnicities have genotypes consistent with those with moderate-to-severe AATD-related lung disease[2,3] We here at Volv Global SA supported by Takeda developed a prediction model to identify symptomatic patients of different races and ethnicities with likely risk of AATD using claims data from a large US database. Our Conclusions The Asian, Black or African American, and Hispanic or Latino populations were underrepresented in the AATD-diagnosed patient pool when compared to actual US demographics. Although racial and ethnic biases were evident, the prediction model accurately identified patients with likely undiagnosed AATD earlier in their diagnostic journey in these subpopulations. Among patients diagnosed with AATD: • There were fewer treated than tested individuals • Treatment rates were lower in non-White and Hispanic patients Evaluating by race and ethnicity, a larger proportion of non-White and Hispanic or Latino patients with likely undiagnosed AATD had asthma. #healthequity #equity #rarediseases #volv #inTrigue #innovation