Algorithms and machine learning provide a new perspective
for development of medical science and practice
We talk about it with Antuan Angelov, Manager Business Development of Cloud Services and Database at KONTRAX. Curious, seeking for innovative approaches and advanced ideas – let me present you Antuan Angelov, who has joined the team of KONTRAX 3 years ago and is always ready to tell us something new and interesting. This year he has started his doctoral studies about Artificial Intelligence Systems in Sofia Technical University. He graduated his secondary education in PHSEA “Kirov”, Robototechnics specialty, and diploma graduation work “Real Time Clock for Apple”.
Neural nets from the humans science to the responsible attitude regarding the capabilities of the artificial intelligence
The development of the medical science is the basis for realization and development of on entire section in the artificial intelligence (AI) – neural nets and machine learning, which are used for achieving specific goals in the medicine.
Today, the neural nets are more and more thoroughly studied. In contrast to the algorithms approach in AI, the use of neural nets is based on training algorithms in way of examples, which is done for trying to shape their biological equivalent – human brain. The Deep Learning is a method for implementation of machines training, where they “learn” regularities in large unstructured data. On of the widely distributed algorithmic approaches are the neural nets – a structure that actually imitates the connections in the human brain. While the neural nets process data, the net of connected elements setups itself and gives an opportunity for the neural nets to interpret the incoming future data.
Public sectors like Health Care, Education and Development of Smart and Connected Cities are directly related to the development of the artificial intelligence and can best explore the benefits in improving people’s quality of life. In this article we will try to present the capabilities of the intelligent machines and their potential in facilitating and improving the care for people’s better health. And in particular for the AI section – neural nets used for the self-training of intelligent and expert systems.
Michael Dell, founder and chief executive officer of Dell Technologies, is an optimist when talking about AI. Mr.Dell likes the idea of humans working together with the machines. "I love the idea about the partnership between people and machines because it was always very clear in my mind", says Dell. We are talking about human beings from one side, and the machines, from the other side. He thinks that the world is divided in people and machines but actually it is a matter of common work.
In principle, using neural nets we can calculate every calculable function, i.e. using neural nets we can do everything that could be done using the standard digital computers – to perform basic binary operations (with 1 and 0). However, in contrast to the algorithmic approach the neural nets use the principle of parallel data processing and therefore they can be successfully used for tasks related to processing of big unstructured data (signals, speech, images, etc.) and that in real time, where we aim at achieving quicker actions. Thus, the neural nets take their power out of the branched processing of information and of the capability for self-training for generating a summary.
The artificial NN facilitate things and try to cover the higher and more abstract features of neuron calculations. The basis on one neural net is the technique for data processing inspired by the way it is done in the human brain. The artificial neural net is a mathematical model consisting of set of separate elements that stimulate some monitored features of thee biological neural systems (as well as the processes of adaptive biological learning of new knowledge and skills). The modern model of neural net is a composition of well interacting elements and their connecting channels.
Machine learning benefits for creation of new innovative approaches in medical practice
In the public health services the AI methods are used mainly in the process of identification and diagnostics of diseases. These techniques help the doctors to make better decisions based on the experience accumulated and brought out of the machines as result of the analysis made on data. The AI may support the medical team in preparing the necessary conclusion by suggesting ways for improving patient’s way of life, prescription of necessary medicines, ensuring automated monitoring, and early assessment of critical conditions, as the expert systems may detect even the earliest and weak signals, which a person could not find immediately.
Dell EMC has a leading position in regard to AI providing technology that makes tomorrow possible today. Dell EMC unambiguously offers technologies – including workstations, servers, network equipment, storage, software, and services for solutions for data analysis, which are in the basics of the AI, the machine learning. Furthermore, Dell EMC invests in creation of portfolio of ready solutions, which with the help of experts’ team provides a possibility for integration and adaptation of every solution for machine learning.
The chatbots integrated in the messaging mobile applications are now even better identifying the patient’s conditions through asking question and during aconversation with the patient, as they compare also the incoming information from the sensors in the remote cloud system for monitoring of that specific patient. They allow for decreasing the workload and investment in the specialized call centres for patients and improve the response time for emergency cases.
Oncology – making the correct decisions. Cancer is a complicated and sometimes indefinable disease, and the unique cancer of every person may result in unexpected challenges. Therefore, the doctors often look for other experts opinions in order to confirm the treatment scheme or to consider new approach based on the supporting proofs, and possibly increasing the trust in their decisions for treatment. Studies show that 10% of the mortality rate caused by lungs’ cancer is related to mistakes in diagnostics, and 4% of the millions of radiation interpretations carried out every year, include mistakes of clinical importance. And just here the neural nets and machine learning may improve these results, but still software development and training are another serious challenge. For example, the expert systems based on AI in the oncology field provide recommendations that “match” with those of the doctors around the world in most of the cases – so, the experts may focus on what they can do better – take care of the patient.
Statistics and risk assessment for disease – the neural net may already predict very accurately what the risk is for death of a patient who has come for treatment at the hospital with advance stage of oncological disease, due to the analysis of earlier non used records in old medical files and PDF files. In total the neural net analyzes hundreds of thousand factors, and that makes her faster and more accurate in comparison to any other currently existing program.
Genetics is another field where AI is increasingly used. It could offer personalized treatment or change in behaviour during the process of treatment. The personalized medicine or the more effective treatment based on the individual health data of the patient combined with prognosis analyses is also a hot field of research and closely related to the better identification of the disease. This scope is currently controlled by a section of AI – machine controlled self-training that allows the doctors to choose, for example, from more limited groups, diagnoses, or to assess the risk for the patient, based on symptoms and genetic information.
Decreasing the time for analysis of genomic researches in parallel with controlling and storing the huge volume of data, including images from different systems (PACS, cardiologic, pathological, etc.), which are generated and used, is the main challenge for the IT infrastructure. Dell EMC Isilon Scale-Out NAS offers highly reliable and secure file system for the applications and workloads. Due to the unique management technologies and smart reduction of the physical space, Isilon is a global leader in the field of solutions for unstructured data.
Invention of new medicines. The machine learning can also accelerate the invention of new medicines. A system based on AI may check more than 10 000 potentially disease-causing compounds. There are artificial intelligence programs developed, that may learn from the earlier successes in choosing the compounds, assessing which of them are toxic for the diseased cells in order to block the action of the parasite proteins and keep the equivalent human protein untouched. Instead of using check methods with numbers, machine learning is used for developing an approach and finding new potential chemical structures.
The technologies, which the AI and machine self-training is based on, are also applied on monitoring and prognosis of epidemic centres around the world, on the basis of data collected from satellites, historical information in Internet, real time updates in social media, and other sources. Supportive vector machines and neural nets have been used, for example, for predicting bursts of malaria, taking into account data such as temperature, average monthly rainfalls, total number of positive cases, and other data. Predicting the seriousness of the centres is very necessary in the Third World countries where often they lack medical infrastructure, the education required, have limited communications, and access to treatment.
Intelligent electronic health records. Classification and categorization of documents (for example: sorting of patient’s requests sent by e-mail), using vectors supporting machines and optic recognition of symbols (transformation of italic or other sketched handwriting into digitalized signs), are main categories in AI, where techniques of machine self-training are used continuously. They provide for the progress in collection and digitalization of information for the electronic health care. The technologies for optical recognition of handwritten sings of symbols are only one of the examples for innovation in this field.
Of crucial importance for the qualitative processing and analysis of data from the electronic records, combined with reliable protection and fast reaction, is the storage where the data stays. Dell EMC XtremIO is a all-flash storage with unique technologies for using less space, exceptional speed of read and write operations, guaranteed security, and consistency of data, that is used in the organization of intelligent electronic health records.
Improving the radiation therapy. Diagnostic radiation in way of added reality and machine reading of images. In the next 20 years the radiologists will not be the same. They will be more like cyborgs who track algorithms reading thousands of researched in a minute.
Sensors and smart devices used for collecting of data about the patients’ conditions. In the next decade another filed that is very quickly developing, will be related to the increased use of bio-sensors and smart devices, as well as more sophisticated options for measuring the health condition and for remote monitoring, which will ensure common unified data flow that can be used for supporting the efficiency of the research and development activity and the treatment. This type of personalized treatment has important consequences for the individual in regard to health optimization, but also for reducing the total costs for health care.
Completed projects for better life quality
KONTRAX as a long-term developer of software for health care also operates for ensuring a remote mode of medical service for patients. Tools with different sensors for monitoring patient’s conditions, including video consultation, are part of the telemedicine mode. In reply to the increased demand for virtual service, the system integrator KONTRAX, is able to provide the medical service suppliers possibility for online service of their patients.
Taking into consideration the necessity and benefits from the implementation of remote medical and social services system in Bulgaria, in 2016 KONTRAX AD has established one of the first systems of this kind for Septemvri Municipality, under the Project №BG05M9OP001-2.002-0277 „Independent life in Septemvri Municipality” in relation to Procedure for direct provision of free financial aid №BG05M9OP001-2.002 “Independent Life” with the financial support of the European Union through the European Social Fund.
The development of similar intelligent systems will ensure that:
- Information for the health conditions of themonitored patient, is available for their relatives aand their general practitioner, within given period of time.
- Innovative ways of treatment can be supported by additional mobile sensors, such as bracelets for pulse measurement, devices for action/ mobility, scales with information about hydration, bone mass and fats, remote pulsimeter for bed, SMART inhalator, air quality detector.
Since the beginning of 2018 Hippocrates GP, a software product developed by KONTRAX for the needs of the general practitioners, includes a new service that automates publishing all medical and diagnostic orders (MDO)via Internet into cloud server. The laboratory apparatuses receive automatically the requests for tests and automatically perform them. The results are validated by the laboratory doctor and are available for reading in Hippocrates GP. The data received from the results of the medical tests are automatically added to the patient’s electronic record that is stored in Hippocrates GP – the health and information system of the general practitioner.
The challenges in front of the AI
The rush for introducing AI technologies in the field of medicine still has its serious challenges that need to be solved, such as:
- Data management is one the most pressing problems that has to be faced recently. The medical data is still private and not easy for access and it is reasonable that most of the people are not willing to share data; it is crucial to focus on the importance of data sharing and integration between branches.
- Simplification of electronic records, which are presently still messy and fragmented in the databases, will be the main initial step of personalized treatment solutions implementation.
Soon a series of articles and materials forartificial intelligence will be published, in which we will provide new ideas and concepts, as well as the current projects Kontrax has developed in the field.
KONTRAX is a leading system integrator in Bulgaria. The company has a wide portfolio of technologies and high expertise in implementation of complex solutions for the health care system. Together with its long-term partner Dell EMC, Kontrax offers computer equipment, servers, storage and network equipment, which are the basis for successful adoption of artificial intelligence in health services.