Dataset Information

Improving genomics-based predictions for precision medicine through active elicitation of expert knowledge.

ABSTRACT: Precision medicine requires the ability to predict the efficacies of different treatments for a given individual using high-dimensional genomic measurements. However, identifying predictive features remains a challenge when the sample size is small. Incorporating expert knowledge offers a promising approach to improve predictions, but collecting such knowledge is laborious if the number of candidate features is very large.We introduce a probabilistic framework to incorporate expert feedback about the impact of genomic measurements on the outcome of interest and present a novel approach to collect the feedback efficiently, based on Bayesian experimental design. The new approach outperformed other recent alternatives in two medical applications: prediction of metabolic traits and prediction of sensitivity of cancer cells to different drugs, both using genomic features as predictors. Furthermore, the intelligent approach to collect feedback reduced the workload of the expert to approximately 11%, compared to a baseline approach.Source code implementing the introduced computational methods is freely available at https://github.com/AaltoPML/knowledge-elicitation-for-precision-medicine.Supplementary data are available at Bioinformatics online.

SUBMITTER: Sundin I

PROVIDER: S-EPMC6022689 | biostudies-other | 2018 Jul

REPOSITORIES: biostudies-other

ACCESS DATA

Publications

Improving genomics-based predictions for precision medicine through active elicitation of expert knowledge.

Sundin Iiris I Peltola Tomi T Micallef Luana L Afrabandpey Homayun H Soare Marta M Mamun Majumder Muntasir M Daee Pedram P He Chen C Serim Baris B Havulinna Aki A Heckman Caroline C Jacucci Giulio G Marttinen Pekka P Kaski Samuel S

Bioinformatics (Oxford, England) 20180701 13

<h4>Motivation</h4>Precision medicine requires the ability to predict the efficacies of different treatments for a given individual using high-dimensional genomic measurements. However, identifying predictive features remains a challenge when the sample size is small. Incorporating expert knowledge offers a promising approach to improve predictions, but collecting such knowledge is laborious if the number of candidate features is very large.<h4>Results</h4>We introduce a probabilistic framework ...[more]

PMID: 29949984

Similar Datasets

Project description:Precision medicine holds great promise for improving health and reducing health disparities that can be most fully realized by advancing diversity and inclusion in research participants. Without engaging underrepresented groups, precision medicine could not only fail to achieve its promise but also further exacerbate the health disparities already burdening the most vulnerable. Yet underrepresentation by people of non-European ancestry continues in precision medicine research and there are disparities across racial groups in the uptake of precision medicine applications and services. Studies have explored possible explanations for population differences in precision medicine participation, but full appreciation of the factors involved is still developing. To better inform the potential for addressing health disparities through PM, we assessed the relationship of precision medicine knowledge and trust in biomedical research with sociodemographic variables. Using a series of linear regression models applied to survey data collected in a diverse sample, we analyzed variation in both precision medicine knowledge and trust in biomedical research with socioeconomic factors as a way to understand the range of precision medicine knowledge (PMK) in a broadly representative group and its relationship to trust in research and demographic characteristics. Our results demonstrate that identifying as Black, while significantly PMK, explains only 1.5% of the PMK variance in unadjusted models and 7% of overall variance in models adjusted for meaningful covariates such as age, marital status, employment, and education. We also found a positive association between PMK and trust in biomedical research. These results indicate that race is a factor affecting PMK, even after accounting for differences in sociodemographic variables. Additional work is needed, however, to identify other factors contributing to variation in PMK as we work to increase diversity and inclusion in precision medicine applications.

Project description:Infection with the Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2) induces the coronavirus infectious disease 19 (COVID-19). Its pandemic form in human population and its probable animal origin, along with recent case reports in pets, make drivers of emergence crucial in domestic carnivore pets, especially cats, dogs and ferrets. Few data are available in these species; we first listed forty-six possible drivers of emergence of COVID-19 in pets, regrouped in eight domains (i.e. pathogen/disease characteristics, spatial-temporal distance of outbreaks, ability to monitor, disease treatment and control, characteristics of pets, changes in climate conditions, wildlife interface, human activity, and economic and trade activities). Secondly, we developed a scoring system per driver, then elicited scientific experts (N = 33) to: (a) allocate a score to each driver, (b) weight the drivers scores within each domain and (c) weight the different domains between them. Thirdly, an overall weighted score per driver was calculated; drivers were ranked in decreasing order. Fourthly, a regression tree analysis was used to group drivers with comparable likelihood to play a role in the emergence of COVID-19 in pets. Finally, the robustness of the expert elicitation was verified. Five drivers were ranked with the highest probability to play a key role in the emergence of COVID-19 in pets: availability and quality of diagnostic tools, human density close to pets, ability of preventive/control measures to avoid the disease introduction or spread in a country (except treatment, vaccination and reservoir(s) control), current species specificity of the disease-causing agent and current knowledge on the pathogen. As scientific knowledge on the topic is scarce and still uncertain, expert elicitation of knowledge, in addition with clustering and sensitivity analyses, is of prime importance to prioritize future studies, starting from the top five drivers. The present methodology is applicable to other emerging pet diseases.

Dataset Information

Improving genomics-based predictions for precision medicine through active elicitation of expert knowledge.

Publications

Improving genomics-based predictions for precision medicine through active elicitation of expert knowledge.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets