Showing 8 ideas for tag "machine-learning"
(@jsmoller) kudos icon +

AoU Research Priorities Use Cases

How can we predict and prevent suicide and suicidal behavior?

Suicide is a leading cause of death and rates have not decreased in decades. Currently there are no established risk factors that have demonstrated usefulness in predicting individual risk. The All of Us program is in a position to change this by leveraging the large and varied data that will be collected. Making progress in this area would be a major advance and could have a tremendous impact on the suffering that individuals... more »

Voting

28 votes
Ideate
(@nkeshava) kudos icon +

AoU Research Priorities Use Cases

Can machine learning algorithms reliably identify distinct sub-populations and sub-phenotypes in a specific disease cohort?

Identify a cohort with a specific medical condition, intervention, and outcome and use heterogeneous patient measurements. Using a novel, soon-to-be-published machine learning platform that performs "X-rays" of cohorts, discover distinct sub-populations, their sub-phenotypes, and different outcome rates. The ability to discover distinct sub-populations characterized across many modalities would greatly advance the... more »

Voting

14 votes
Ideate
(@alexandra.boukhvalova) kudos icon +

AoU Research Priorities Use Cases

Differentiating healthy neurological aging from incipient nervous system dysfunction using smartphone apps.

Neurons are mostly non-renewing cells and neurodegenerative diseases are often diagnosed after many already died. This severely limits therapeutic efficacy. We propose accurate, convenient measures of broad neurological functions using developed smartphone apps to test whether derived subject-specific progression slopes can differentiate healthy aging from incipient neurological dysfunction. Such analysis may eventually... more »

Voting

24 votes
Ideate
(@andrew.cocke) kudos icon +

AoU Research Priorities Use Cases

What is the relative impact of various physical environmental factors like natural daylight, view, and noise on patient oucomes?

Using data on the physical features of patients' rooms, bathrooms, nurses' stations, etc., this study would use multivariate linear regression, artificial neural networks, deep leaning, or other AI methods to draw correlations between each of these physical features and patient outcomes like length of stay, readmission rates, accidents, hospital-acquired infections, mortality, and staff occupational injuries, and build... more »

Voting

13 votes
Ideate
(@skbandt) kudos icon +

AoU Research Priorities Use Cases

Can we advance our understanding of epilepsy by developing new, data-driven tools for use in its diagnosis and treatment?

Gather MRI, genetic and basic clinical and behavioral data from all patients with epilepsy (conservatively 1% of the population). Data can then be analyzed using a machine learning approach to identify consistencies and differences among patients which can then inform our understanding of epilepsy subtypes and subsequent treatment decisions.

Voting

11 votes
Ideate
(@lhwostow001) kudos icon +

AoU Research Priorities Use Cases

Conduct research to determine if blood can help identify the presence of the HPV-16 virus.

Goal: save millions of lives. A person dies from oral cancer every hour of every day.
Methodology:
- Include people diagnosed with oral and oropharyngeal cancer caused by the HPV-16 virus
- Collect and analyze blood every 3 months within the first 2 years of diagnosis. Between 2 and 5 years do it every 4 months
Outcome: increase the survival rate. The death rate is about 43% at five years from diagnosis. When found at... more »

Voting

3 votes
Ideate
(@visarb) kudos icon +

AoU Research Priorities Use Cases

Can measures of speech and language be pre-clinical predictors of neurological disease?

Complex brain networks are responsible for translating thoughts and ideas into sequences of movements that produce the sounds of speech. Any disturbance in these networks can lead to changes in how we speak and what we say. Evidence shows that subtle changes in speech are early harbingers of neurological disease. We propose to use the AoU platform to validate this hypothesis on a large scale and to develop AI tools for... more »

Voting

2 votes
Ideate
(@kjmaier) kudos icon +

AoU Research Priorities Use Cases

AoU’s sample/measures uniquely allow for modeling high-burden disease pathways specified by the biopsychosocial ecological model

Biological, psychological & social domains interact at multiple levels (genes, human & environment microbiomes, stress, behavior, broad SES & environment) outlined by the biopsychosocial ecological model (BPSEM). The BPSEM specifies pathways aligned with the AoU pillars of biology, lifestyle, & environment. Simultaneous examination of the interacting domains and levels of the BPSEM is now computationally possible and... more »

Voting

1 vote
Ideate