Cracked the Code: Now What?

July 17, 2017

By: Stephanie Allen

The simplest definition of Omics is a field of study in biology ending in – omics. Some of the more familiar are genomics, proteomics, and metabolomics. Combining experimental biology with computational biology has spurred major advances and drives the many disciplines that fall under the larger umbrella of Omics. One of the most important advances has been the reduced cost in genome sequencing. The genome however is not alone in the cascading landscape of data. The epigenome, microbiome, proteome, metabolome and transcriptome are a few areas of Omics that are getting attention. As they should. After all, it’s not just DNA writing life’s rules. If we consider the above mentioned, then it is the influence of our environment (epigenome), the bugs we live with (microbiome), the motors that run the whole thing (proteome), the chemicals of life’s party (metabolome) and a messenger to interpret what’s in the DNA dictionary (transcriptome). Each colossal on its own and some would argue not inclusive of all “Omics” that have some influence over life.

The advances that have been made in the many areas of Omics are fascinating. The completion of the human genome project in 2003 was the most important milestone in biology in the last few decades. Machine learning is now being layered on to solve some of the most pressing problems in healthcare. One such example is data collection in the ICU. If you have ever stayed in the hospital for even one day you know the routine of constant monitoring. Unfortunately, most patients in the ICU must stay longer than one day. The breadth and frequency of data collection in this scenario is staggering when you think about every patient in every hospital. This life saving data is necessary to optimally diagnose, treat and discharge patients and is also an opportunity to collect valuable data sets for research.

Applications in machine learning are now tackling one of the major issues in ICU data collection. Namely, false alarms that are known to create an unsafe environment for patients. In part, because caregivers become desensitized and life threatening events can be missed (1). Layering on machine learning to patient data collection has shown to significantly improve quality of alarm signals. Machine learning algorithms can aggregate and analyze a variety of physiologic data measurements such as carbon dioxide, blood pressure, heart rate and oxygen saturation all at the same time and in addition, suppress alarms when there is a physiologically implausible reading (23). Machine learning can help overcome issues caused by monitoring equipment that are often manufactured to have extremely sensitive alarm thresholds which leads to a disproportionate number of false alarms observed in the ICU. Improving acute care is an important priority and has already shown promise.

Beyond acute care, we know that ICU patient data sets are an incredible opportunity for biomedical research. Indeed, the large data sets are an opportunity, but also present many challenges. Simply how to make the data accessible is a bottleneck. The creation of openly available databases such as MIMIC (4) that houses de-identified patient records for over 30,000 patients is a key step towards making patient data available to the public a more common practice.

Applications of machine learning in acute care is a good start and the right place to start. However, it is only the beginning. Consider the many areas mentioned above that fall under the bigger umbrella of Omics. The scientific advances will be widespread and will no doubt make a huge impact. Especially now that machine learning and AI have come to the the biology playground. Certainly we will see healthcare change to become more personalized with precision medicine. Better diagnostics and treatment strategies are already on the rise. The insurance industry will also have to adjust. There will be more data available on clients that could hinder their insurability but with making the data available for more precision treatment plans the insurance industry could gain huge savings. To name a few of the less obvious, Omics has the potential to improve agriculture, robotics, and when you think about it even the entertainment industry. And now it’s your turn, what do you imagine is on the horizon?