Artificial Intelligence is one of the technologies that is leading the digitization of Health. The health ecosystem is taking advantage of the advances that exist on this front, and there are quite a lot of interesting projects that are applying it successfully. In this article we are going to approach real cases of the AI for Health: use cases through APIs, which also allow their rapid integration into third-party software.
Tabla de Contenidos
- 1 HOSPITAL MANAGEMENT: NATURAL LANGUAGE PROCESSING
- 2 IMAGE DIAGNOSIS: IMAGE RECOGNITION AND MACHINE LEARNING
- 3 DIAGNOSIS SUPPORT: NATURAL LANGUAGE PROCESSING, BIG DATA AND DEEP LEARNING
- 4 INTELLIGENT REMOTE MONITORING: IOT, BIG DATA AND DEEP LEARNING
- 5 CLINICAL TRIALS: NATURAL LANGUAGE PROCESSING, MACHINE LEARNING AND BIG DATA
- 6 CONCLUSIONS
HOSPITAL MANAGEMENT: NATURAL LANGUAGE PROCESSING
The Clinical History of each patient hides a huge amount of information.
Thanks to the digitalization of the Clinical History, an important change is taking place in two aspects:
- The reports and tests results are no longer stored on paper, but in pdf or similar documents.
- The physical folders that keep the traditional Clinical Record are being digitized by scanning the documents.
As a result, the amount of Electronic Clinical History is increasing with respect to the Clinical History on paper.
However, it is extremely difficult to digitize habits: most health professionals continue to report their notes, follow-ups and diagnoses by writing free text, that is, unstructured.
And unstructured data is non-processable data.
Therefore, we still have a lot of information of enormous value, hidden in the patients’ Clinical Histories.
How do we make all that information surface so that we can process it?
The Natural Language Processing is a technology belonging to the field of AI that consists of processing text written by a human being. Among other actions, they are algorithms capable of filtering words that do not provide meaning, identifying words with meaning, compound expressions, labeling relevant terms, either by frequency of use, or by contrast with an external knowledge model, etc.
Applied to a medical report, and supported by medical standards and ontologies such as SNOMED-CT, CIE-10, etc., it allows extracting problems, symptoms, diagnoses, medications, specialties, etc. from the free text notes that contain those reports.
In this way, it is possible to automatically extract structured information from each report.
This has an enormous value for the healthcare center, allowing it to optimize the management of the agendas, optimize patient follow-up, and a long list of other actions that improve clinical management in general.
Imagine what you could do if your software could integrate a fully mature and reliable NLP solution using an API.
IMAGE DIAGNOSIS: IMAGE RECOGNITION AND MACHINE LEARNING
Image recognition has evolved a lot in recent years, especially in combination with Machine Learning.
This type of solution requires training. The specialist loads images into the system, and the system issues a result based on its history and certain initial variables.
In the first iterations, the system shows a high degree of error, because it lacks historical and faces a high degree of uncertainty in these first images.
The specialist supervises the training process through a labeling system that the software uses to adjust its criteria variables, which together with the previous results, quickly improves its success rate.
With an adequate training, this type of solutions reach spectacular levels of precision, similar to the human specialist.
The application of this technology in the support of Diagnostic Imaging is enormously important, especially in techniques of radiological study by layers, or that require a high degree of visual precision.
There are highly specialized solutions that can be integrated into clinical systems and Apps to help the user identify possible injuries automatically, and with a high level of reliability.
Having these solutions through APIs greatly simplifies this integration, and multiplies the possibility of putting them in the hands of a greater number of health professionals.
With this technology, diagnosis time is significantly reduced, which has direct impact on healthcare improvement.
DIAGNOSIS SUPPORT: NATURAL LANGUAGE PROCESSING, BIG DATA AND DEEP LEARNING
We have discussed the application of NLP to the improvement of clinical management. But there are other applications.
The extraction of clinical data from medical reports allows a huge quantity of information to be stored in a processable way.
For example, a set historical data on symptoms, patient profiles, treatments, diagnoses and evolution, which opens the door wide to other satellite branches of AI, squeeze all that information and feed intelligent systems.
Big Data and its ability to analyze massive amounts of information can be the engine of a diagnostic support system. This type of systems, starting from a previous training through Deep Learning solutions, can reach brutal levels of precision because the daily use of the health professionals supposes a constant additional training that would improve even more the rates of precision and accuracy.
INTELLIGENT REMOTE MONITORING: IOT, BIG DATA AND DEEP LEARNING
In recent years, the patient empowerment and the orientation to preventive healthcare are two of the main strategic lines in healthcare.
In both cases, AI plays a fundamental role.
There are several technologies involved in this use case:
- IoT (Internet of Things), capturing data from the patient, their environment and their routines, in a “raw” state.
- Deep Learning, correlating that data, and building rules of patient behavior that allow the system to learn to “know” their specific routines.
- Big Data, analyzing a quantity of data that is constantly captured to feed the knowledge base of the system and improve its accuracy.
This type of solutions allows to generate alerts when the patient moves away from his habitual routine, or when in his environment something that supposes a potential risk occurs, etc.
Thanks to these alerts you can act before the patient is affected, or with a speed that in many cases can be vital.
Integrating this type of systems through APIs allows these advanced monitoring capabilities to be available to more and more care and chronic patients care centers.
CLINICAL TRIALS: NATURAL LANGUAGE PROCESSING, MACHINE LEARNING AND BIG DATA
Another one of the many doors that opens the extraction and structuring of the information contained in the Clinical Histories, is that of the Clinical Trials.
When a Center transforms its unstructured reports into structured data, it exponentially multiplies its capacities for information analysis and decision making.
In the specific field of Clinical Trials, this structured information allows the Center to greatly optimize the detection and matching of participants, not only quantitatively, which impacts on the goodness of the results, but also qualitatively, by having accurate and easy to correlate information.
When these solutions also combine the NLP with Machine Learning algorithms, they also ensure optimal accuracy in the automation of the entire process.
The result is better Clinical Trials, and greater capacity to perform more Trials every year, in the same Center or in multiple centers.
The extension of this type of solutions also allows to increase exponentially the capacities of the entire ecosystem, since it leads to building a Big Data of a brutal value for clinical practice in general, and for Clinical Trials in particular.
Integrating this type of solutions through an API allows us to join forces in a network of information of high value for the health ecosystem and, therefore, for the whole society.
Beyond the most mediatic applications of the AI, such as autonomous cars, virtual assistants, etc., the application of its different technologies in Healthcare is a reality that the APIs are helping to spread rapidly.
This ability to easily integrate advanced NLP solutions, Image Recognition, Machine Learning or Deep Learning, Big Data, etc., is key so that all the advantages they bring to the health ecosystem are extended as much as possible.
In this way, we improve among all the quality of life of citizens through a more advanced and efficient Health:
- Systems for clinical diagnosis support
- Diagnostic imaging support
- Improvement of Hospital Management
- Remote Monitoring of patients
- Clinical Trials
All these areas and many more, are improved drastically thanks to the use of these technologies, available through APIs for their rapid integration into the software used daily by millions of professionals and citizens worldwide.