Due to the importance of healthcare to the humanity and the amount of money concentrated in the industry, its representatives were among the first ones to see the immense benefits they can gain from innovative data science solutions. For healthcare providers, it’s not just about lower costs and faster decisions. Data science also helps provide better services to the patients and makes doctors' work easier. But that’s theory and today we’re looking at specifics.
Health Information Systems
Health information systems collect patient data and compile it in a way that can be used to make fast and accurate healthcare decisions. In hospitals, Clinical Decision Support (CDS) software analyzes medical data on the spot and provides advice for practitioners. For example, if a patient’s blood pressure increases too much, the system will send a real-time alert to their doctor.
And since there’s plenty of money in the healthcare industry, numerous startups develop systems for individual usage outside hospitals. These products use patient’s data in conjunction with data from CDS in order to create better treatment plans for people with asthma, diabetes, and other diseases. The algorithms based on deep-learning increase the diagnostic accuracy by analysing previous examples and then suggest better treatment solutions.
Hospitals collect patient’s digital medical history called Electronic Health Record (EHR) (or Electronic Medical Record, EMR). EHR includes demographics, medical history, treatments, and test results. It can send out reminders when a patient needs to get a new lab test or track if a patient follows doctor’s orders.
Integrating EHR with CDS presents great opportunities to improve patient diagnosis and reduce visits to hospital as well as be used for preventing diseases. For example, EMR was used to predict myopia progression in East and Southeast Asia which affected an estimated 80%–90% of high school graduates. The algorithm was able to predict which children were at high risk of getting sick. This way preventative measures could be taken and the epidemic could be stopped.
Drug Discovery
Producing new drugs costs billions of dollars and takes years of time (twelve years on average to get a drug officially submitted). Machine learning algorithms and AI already simplify this process by reducing the number of required lab experiments for metabolic-disease therapies, cancer treatment, immuno-oncology drugs, etc. Time reduction is especially important in a case of acute epidemiological situation.
Medical Image Analysis
Medical equipment (X-ray CT, MRI/fMRI, etc.) produces complex and highly informative 2- and 3-dimensional images. Analysis of big volumes of raw visual data is extremely time consuming. Data science became one of the major tools of medical image analysis.
Machine learning methods, content based medical image indexing, and wavelet analysis for solid texture classification are used to detect tumors, artery stenosis, organ delineation, etc. It is important to note that systems based on machine learning detect diseases not only faster but also more accurate than humans. For example, an algorithm, called CheXNeXt, outperforms radiologists in diseases detection using chest x-rays by 33% in accuracy.
Applications of Big Data in “-omics”
The organic and molecular fields, such as genomics, proteomics, metabolomics macrobiotics, and other types of “omics”, collect significant datasets. Data science techniques allow integration of different data types to realize the strategies of diseases treatment and drug response. This type of approach enables advanced treatment personalization.
There are already more than a hundred applications that use genomics (MapReduce, Clue Go, GSEA, Pathaway-Express, etc.), proteomics (MaxQuant, Mascot, SEQUEST, and many others) to help in preventing or curing diseases.
Microsoft HoloLens
This technology is an example of using AI in surgery. HoloLens combines elements of virtual reality and augmented reality to map the anatomy of a patient and overlay 3D images of blood vessels, bones, and organs on top of their limbs. This pre-operate procedure helps to “see” the unique anatomy of each patient and plan the operation accordingly.
Conclusions
The usage of healthcare data analytics in medical institutions is growing extremely fast. Main reasons for that are:
- High volumes of data which require real-time analysis.
- Handling complex data needs special analytical techniques.
- Data science provides efficient data collection technologies.
Data science techniques in the healthcare system are going to play a vital role in how healthcare is practiced in the future.
