Machine Learning for Healthcare Sector

“Technology disrupts many jobs but it creates new jobs with more efficiency and consistency to become the lifeline for the economic growth of any country”.

In today’s internet-driven world, IR4.0 revolution is riding on the high wave of technology and information. Consequently, many events are impacting our lifestyle, health, relationships, productivity and efficiency. In such a dynamic environment, you are required to do multi-tasking while dealing with management, customers, vendors and business partners. You should be physically and mentally fit to endure such a high-pressure environment and get affordable healthcare when needed. Digital Technologies, Intelligent Sensors, and Multi-Media intelligent devices are already supporting hospitals to provide good healthcare. Emerging technologies like Artificial Intelligence (AI), Machine Learning (ML), Drones, Robotics, Cloud Computing, Big Data, Internet of Things (IoT) and 5G (future) are moving fast to reshape the work practices and business processes of many industrial sectors. The demand for AI / ML systems and skilled professionals to support AI/ML applications continues to rise. During 2019, some industry analysts had estimated that the AI/ML Healthcare market may reach $6.6 billion by 2021. Due to COVID-19 pandemic, these estimates may be revised.

Machine Learning is a subset of AI where it can quickly access large data from healthcare database to speedily process, diagnose and interpret accurately. The outcome can assist the physicians to take the final decision. ML is well suited for accurate and fast diagnostics, related to chronic diseases.

Machine Learning (ML). ML is an efficient and very fast computing system, interconnected with sensors, data channels and Big Data system. ML can quickly learn and adapt to the designed model based on healthcare data. ML follows adaptive technique through iterative processing for quick learning. It diligently processes the received/retrieved healthcare data to provide a consistent and accurate outcome. ML can also use data received through multi-media devices like smartphones, in the form of text, image, audio, video and graphics. ML is a highly progressive approach for the precise, cost-effective and comprehensive diagnostics in most of the Clinical Fields such as Oncology, Cardiology, Orthopaedic & Trauma, Urology, Nephrology Plastic Surgery, Neurology, Endocrinology, Genetic Related Disorders and other cases of Mental Health diseases.

Healthcare data. It is quite huge and of different formats /forms. You need special software tools like Data Mining and ML to extract relevant data and process for drawing an inference. All these methods are used to extract target (labelled) data, and present that for processing and determining patterns and relationships. This helps in decision- making.

Neural Networks (NN). NN as a part of the ML system attempts to mimic the way the human brain tackles health problems. In this, ML uses layers of interconnected units to infer relationships based on the observed data of the patient and/ or the data retrieved from the database.

Natural Language Processor (NLP). NLP processor helps in processing large amounts of natural language data (Voice and Text). NLP helps in natural language operations, automated speech recognition and automatic conversion of speech to text.

The popularity of ML. ML is related to statistical data of patients/diseases, and can quickly process data using high computing power and storage of Cloud Computing. ML processes the data iteratively to draw timely and reliable outcome. By implementing ML system, the Hospitals and Pathological Labs will acquire the ability to continually predict changes in their processes/procedures and predict future trends. The outcome of the ML system will be speedy. accurate and unbiased to provide a patient-centric diagnosis. This will help them to update their operations continuingly.

Indian National Strategy on ML for Healthcare. In 2015 India had taken a very bold imitative “Digital India”. To give further impetus to its digital economy, NITI Aayog (Planning Commission) of India had formulated comprehensive guidelines in Jun 2018, for introducing AI, ML and Robotics in various sectors, including healthcare. There is an acute shortage of doctors and nurses in Indian hospitals. India has also an issue of long-distances between its Primary Health Centers (PHCs) located in far-flung areas and referral hospitals located say 200Kms -1000Kms away in cosmopolitan cities. Tele-medicine using ML technology could mitigate this distance problem. ML combined with the Internet of Medical Things (IoMT) will be the new mantra for planning and providing quality healthcare at affordable cost and to a greater satisfaction to the patients.

Indian Healthcare System.  “Health is Wealth” is ever true for any economy to grow for which it needs a healthy and skilled workforce. Therefore, maintaining the good health of a nation is the prime responsibility of every government. Healthcare relates to providing efficient diagnostic facilities and appropriate treatment for the patients so that they remain healthy to work at their full potential. In today’s era of technology and fast mobile communications, we keep changing our lifestyle and habits related to food, drinking, sleeping, and even social interaction. Likewise, our work environment is also fast-changing, where we are doing less physical exercise and eat fast/processed (trash) food, which lacks balanced nutrients. Consequently, we are becoming more prone to many injuries, diseases, viruses and Covid -19 like a pandemic.

Despite India’s rapid progress in many fields, and having good numbers of medical colleges, it still has only a few senior/ experienced physicians in tier 2 and tier 3 hospitals. The doctor on duty in a hospital has to depend on a lab test and analysis report prepared by the lab technicians /junior doctors. Duty doctor has little or no time to refer to a patient’s historical data or any statistical information on similar medical cases of the past. The situation is really poor in Indian government-run hospitals at district/tehsil/ division level. In serious cases, PHCs located in a far-flung area, just transfer their patients to higher-level hospitals or Research Institutes. However, higher-level hospitals are often quite far away. which causes an undue delay in treating the patient.

Patient Data Types. Patient’s data related to his/her current and past ailment and data of patient’s relatives /family members and other patients having suffered from similar disease can be speedily collated and processed/interpreted by the ML system. The outcome is then presented to the senior doctor to decide the final procedure/ treatment. Some common data formats and the medium used for recording images and graphic data about patients and which can be integrated into the ML system are listed below:

·        Patient personal details. Name, Unique ID (Addaar Card in India), Next of Kin (NOK). Contact phone, address. Direct admission (D) or Transfer case (T) from other hospitals.

·        Disease Data. Patient Name. hospital registration number, ward number, bed number, date of admission, time of admission, age, gender, disease name and code (WHO Code).

·        Medical Record Sheet. Handwritten case sheet or hand-held keypad for on-line keying in digital data for each patient.

·        On-line monitoring in ICU. Health parameters like BP, Pulse-rate, Heart-beat and body temperature.

·        Blood test data. Liver profile, Blood Sugar. Haemoglobin. Blood count, Platelet Count. Cretin, Uric Acid

·        Heart Test. ECG, Angiography, Doppler Ultrasound test

·        Image Scans. X-Ray, Ultrasound Scans, Magnetic Resonance Imaging (MRI), Computed Tomography (CT) Scan, Brain Scan.

·        Breast Cancer. Mimeograph Scan

·        Microscopic plates. These are sample data of Cancer, Tuberculosis, and Viruses.

Sources of Healthcare data. Healthcare data mainly comes from Research and Development Labs, physicians, hospitals/clinics, patients and social media. It is also extracted from the database held on a chosen cloud.

Algorithm Development. The heart of the ML system is diligent designing of the algorithm and its strict validation by a joint team of experts. Algorithms for healthcare applications are more complex than those used for other business applications. Healthcare algorithms are very carefully developed by a highly skilled and multi-disciplinary team, having in-depth medical knowledge and experience in high-end software development. 

Software tools and Programming Language. To solve healthcare problems, the project team uses a variety of software tools like TensorFlow, NumPy and programming languages like Python and R. For maintaining the database, the team may use MongoDB / MySql like database and Cloud-like AWS (Amazon), Azure (Microsoft), IBM Cloud or Google Cloud. Software team uses numerical methods to transform expressions/ equations (matrices, calculus and statistic) into programmable expressions. Thereafter, Python and R like programming languages are used for codification. Each developed algorithm is first verified by the software team and then validated using test data, jointly by healthcare experts and software team. The formulation, analysis, design and statistical modelling, coding and testing and validation should be treated as a Software Project Management, adopting Agile /Scrum like approach.

Algorithm Selection. Expert / the doctor must carefully select an appropriate algorithm for the selected real-life healthcare problem.

Training of Computing Machine. ML training is an elementary but laborious process like a mother teaching her small child by asking the child to listen carefully – “what to do, when to do and how to do”. She even cautions her child what not to do and make no further mistake. The child grasps the mother’s instructions and faithfully tries to do as was told. If the child makes a mistake, the mother corrects the child but warns him/her to be careful in future. This cycle carries on iteratively to gain more knowledge and respond to any emerging situation. The child also learns to choose alternative action to deal with any new situation. This way the child progressively becomes confident in his/her actions. The mother becomes confident that her child can deal appropriately with any emerging situation. A similar concept is being followed to train the ML system so that it could process the data faithfully, speedily and diagnose the disease accurately. 

Types of ML Training. The outcome of the ML system is the output generated by a well-trained computer/ machine, clean data and validated algorithm. A predictive algorithm will produce unbiased prediction based on the data. ML system can be fully trained on various models and by using various data-sets from the database and even actual patients. Different approaches for ML system training, based on data structure, type, volume and complexity are:

·        Supervised Learning. (SL). It is human supervised learning by the computing machine, based on known data-set attributes and data classification. This data has labelled features that define the meaning of data. The machine is trained using pre-processed examples, and the performance of the algorithms is evaluated with known test data.

·        Unsupervised Learning (UL). It is an iterative process used for analyzing data-set but without human intervention. UL is best suited when the problem requires a huge amount of data coming from social multi-media applications, such as Messenger, WhatsApp, Twitter, Facebook or Telegram.

·        Reinforcement Learning (RL). RL is a behavioural learning a model where the algorithm receives feedback from the analysis of the complex data and the user (computer system) is progressively guided to the best outcome. ML system learns through trial and error but provides a better outcome.

ML Applications in Healthcare. Applications of ML in healthcare is a very vast subject and only general coverage has been given in this article. Incidences of complex diseases are growing at an alarming rate. This is further compounded due to large Volume, Variety and Velocity (V3) of structured/unstructured data flowing through the internet on a 24x7 basis. Fortunately, this issue is mitigated by the availability of Cloud Computing and Big Data technologies. Thus, ML can provide timely, affordable and better healthcare services. ML is the most valuable resource for the healthcare sector and support, many healthcare experts, senior doctors/physicians and pathologists. Luckily, the healthcare sector has been an early adopter of digital technology, optics, instrumentation, intelligent sensors and robotics.

Application areas. ML is an integrated system of many computing devices, data monitoring sensors, supporting high-speed data channels having capability and capacity for data capture from multimedia devices like smartphones/tablets. ML can process retrieved data at super speed, carry out algorithm-based analysis across vast data related to patient/disease and even suggest a treatment. Some areas where ML is best suited are briefly indicated below: -

·        Healthcare data of patients and their relatives and various diseases is too large to handle by traditional analytic techniques.

·        Past historical data of patients and diseases are too large and takes too much time to retrieve from medical records (Paper/ Digital).

·        The urgency of the patient’s current and past data records of chronic diseases like cancer, diabetes, tuberculosis, depression and suggested treatment.

·        Rules/constraints and quantification of disease parameters are not easy to translate.

·        Speed, Accuracy and Interpretation of the outcome are most important.

·        The large volume of structured/ unstructured data (Text, Voice, Image, Graphics) related to patients is flowing through multi-media

Job Potential. During 2019, Frost & Sullivan a Research firm in the USA had estimated that by 2021, AI /ML will generate nearly $6.7 billion in revenue in the global healthcare market. Likewise, McKinsey has predicted that application of Big Data, Cloud Computing and ML in the healthcare sector can generate globally up to $100 billion annually. This will be possible due to the availability of AI/ML, Robotics, Intelligent sensing devices, Cloud Computing, Big Data enabling faster and more accurate decision-making. ML will also offer new software tools for accurate, unbiased and timely prediction for supporting physicians, and drug regulators. There are many ML applications in the healthcare sector. Some of these are briefly given below: 

·        Identifying Diseases and Diagnosis. With the rapid advancement in computing, databases architectures, AI, Robotics, Internet of Medical Things (IoMT), integration with 5G technologies and multi-media devices.

·        Medical Imaging Diagnosis. ML can quickly scan and interpret the contents in the image. For this, ML has helped in developing a new technology called Computer Vision (CV). One such Microsoft project “Inner Eye” is becoming popular.

·        Radiotherapy. During image analysis, there can be many discrete variables which can arise at any particular time. Some of these cannot be easily modelled using complex mathematical equations. Google's DeepMind Health is actively helping researchers to develop ML algorithms to accurately differentiate between healthy and cancerous tissue. This will improve radiation treatment.

·        Accurate detection of Breast Cancer. Breast cancer continues to be a global serious health problem. As per WHO every single year, over 1.6 million women are diagnosed with breast cancer and given treatment. Despite improved treatments, there are globally round  500,000 deaths due to breast cancer every year. Applying ML to mammography screening will help very many women for timely detection and appropriate treatment of breast cancer.

·        Early prediction of epidemic /virus. AI and ML-based technologies are being actively used for monitoring and predicting epidemics around the world. Today, scientists have access to a large amount of data collected as satellite imageries; real-time drone-based data, social multi-media updates and website information. All such data can be quickly collated to predict outbreaks of infectious diseases. Such predictions can be very useful for the WHO to give timely advice to various countries.

·        Data Sharing. With the easy availability of low-cost multi-media portable devices, patient’s data can be speedily passed on to a central database from remote locations.

·        Discovery of New Drugs. With the worldwide spread of the deadly virus “COVID 19”, the urgency to research and find a suitable vaccine to cure corona affected patients have been on war-footing since Jan 2020. A lot of research is on in many labs of hospitals / medical institutes/universities of USA, China, UK, France, Japan South Korea, Israel and India. Discovery and testing of a new drug for human use take a few years. However, to find new treatment against COVID 19, collaborative efforts at WHO level are on and the new drug may be available by early 2021.

·        Drug Research and Manufacturing. ML is already playing an important role in Research & Development of next-generation sequencing and precision medicine. This can help in finding alternative therapy of serious diseases. ML-based software libraries, cloud-based database and fast computation through cloud computing are now available for accelerate drug research work.

·        New Treatment Options. ML the system with real-time patient data available from different healthcare systems of multiple states /countries can increase the efficacy of new treatment options.

·        Remote healthcare. The traditional e-healthcare models based on Information and Communication Technology (ICT), linking PHCs located in rural areas are not adequate to face the challenge of the ever-growing population and onslaught by many deadly diseases and viruses. ICT based healthcare model is required to be integrated into ML systems which will be soon available at main hospitals and research institutes. ML system plays an important role in acquisition, management and exchange of medical information of the patient located in remote/rural area. This will make it possible to follow the treatment/ procedure without moving the patient.

·        Tele-Health. ML has no restriction of the geographical boundary of a district/state/region/country. Thus, ML can mitigate the long-distance the problem of rural PHCs to seek reference from far off referral hospitals and buy medicine locally. Thus, ML will help in providing tale- medicines to a much larger population.  

·        Mental Healthcare. Mental health is an indicator of emotional, psychological, behavioural and social well-being of an individual. It determines how an individual thinks, feels and handles emerging situations. His /her productivity. job progress and family happiness are linked to mental wellbeing. Mental health is an important factor in the various stages of our life starting from childhood to adulthood, women’s pregnancy and old-age. Biological changes in the chemistry of brain functioning lead to hypertension, stress, anxiety and depression. These ailments affect the productivity of individuals, the profitability of the organizations and economy of the country. Therefore, early detection of the onset of mental illness, providing timely and appropriate treatment are very essential. Some issues where ML can support the psychiatrist are:

•  Mental disorders. These include Depression, Anxiety, Bipolar Disorder, Dementia, and Other Psychoses. As per WHO global estimates in 2019, cases of mental illness were - Depression- 264 million, Bipolar disorder – 45 million, other psychoses-20 million and Dementia -50 million. The numbers of mental disorders continue to rise rapidly and it is impacting human wellness, social relationships, human productivity and national economy worldwide. ML can be effectively used for screening patients and assessing the onset and symptom trajectory of the mental illness.
•  Detecting causes of Mental Disorders. Mental disorders may be due to the inability to manage one's thoughts, emotions, behaviour and interaction with others. Mental disorder is caused by biological, social, cultural, economic, political, and environmental factors. Healthcare systems, particularly in developing and under-developed countries have not yet responded suitably to manage their mental health issues. It is envisaged that ML when fully integrated with the domestic healthcare system, could mitigate most of the issues and provide affordable mental healthcare services.
•  Status in India. National Institute of Mental Health and Neurosciences ( NIMHANS) at Bengaluru and New Delhi are already collaborating with local IIIT/ IIT in planning ML-based research and applications for diagnosing and treating Depression,  Hypertension and Anxiety.

Institutes/ Companies engaged in ML Implementation. ML is playing a major role in Clinical Diagnostics, Patient Information System (PIS), early detection of disease, and faster decision making by the healthcare department /organizations/hospitals. There are several institutes/companies in the USA which are actively engaged in ML related activities. These organizations have the potential to offer jobs for professionals with the latest ML related skills sets. Likewise, there are drug manufactures and drug research organizations in technology-oriented countries like USA, UK, Germany, France, Sweden, Canada Japan, Israel, Austria and some emerging economies like India, which are giving a boost to ML use for healthcare.

Advantage of ML for Healthcare Services. Earlier doctors /scientists were using super-computers for analyzing data related to genetic disorders for various diseases of patients. Many times patient’s past records were not available for drawing any co-relation. With the availability of  Cloud Computing, Big Data, Analytics and Multimedia Analytic, ML has now got the ability to quickly retrieve any data and process it at super fast speed. This has given a new dimension for genetic analysis and disease interpretation. Some of the major advantages of using ML for healthcare are given in succeeding paragraphs.

·        ML performance is consistent since the machine does not get fatigued like human beings.

·        ML uses the experience of many senior/ experienced doctors and technical support staff and has a higher probability to be accurate in its prediction.

·         Unbiased, speedy and accurate processing of patient data helps in quick interpretation and timely decision-making.

·        ML can help to discover what genes are involved in specific diseases and also provide a patient’s genetic information.

·         ML can determine which treatment will be most effective for an indi-vidual patient

·        Predict likely outcomes of medical/surgical procedures/ practices and statistical modelling rules.

·        ML can predict waiting/processing time in hospital facilities and do optimal rescheduling of labs testing and procedures in operations theatres.

·        ML can monitor bed occupancy in the number of ICUs and display likely bed vacancy.

·        Detect unexpected behaviour of patients and overcome their depression/anxiety.

·        Sudden quitting of specialist does not affect healthcare services

·        Early detection of disease, its symptoms and causes can help in better administration and control of any epidemic.

·        Shortages of senior doctors/ technicians in remotely located healthcare centres/hospitals can be mitigated by on-line medical advice and medicine.

Issues and concerns about ML. Treating patients using ML is a more delicate procedure than testing ML for an e-commerce application. Indeed, acceptance of ML for healthcare will always be a big challenge due to human emotions, technical constraints and regulatory compliance. The software and algorithm resident in the ML computer will process inputs and generate a program which produces output.  This process goes on iteratively until the outcome is acceptable.

The availability of Big Data supported by the huge computing power of Cloud Computing has given big momentum to ML for decision-making and administration of healthcare services. The output of ML is entirely dependent on the data and algorithm and any change in data will change the outcome. Despite many advantages of ML, some lingering issues for ML use in the healthcare sector are briefly listed below:

·        Human Touch. Some patients feel that machine-based diagnostic lacks the human touch. They believe that a doctor is always sympathetic and cares for his/her patients.

·        Loss of Jobs. Many people in the healthcare sector fear that ML will take away their jobs. They resist ML implementation and avoid learning ML skills even if such free training is organized within their own campus.

·        Lack of trust in the Outcome. In ML environment, it is the algorithm which determines how to process patient’s data and interpret its outcome. They think decision-maker is a black-box and not an able doctor.

·        Data Reliability. Although ML algorithms process input data in an iterative way to produce predictable and accurate outputs based on the data patterns yet there are doubts about the reliability of the data and the outcome.

·        Lack of standards. Although ML has been around for a few years yet there are no universally accepted standards to improve its universal acceptability.  

·         Safety and security of data. Some patients have fear of cyber hackers sabotaging  ML  process to produce the wrong outcome.

·        Ethical implications.  Patients are worried about misuse of their health data by the multimedia service providers.

Future ML Applications. There are lots of opportunities for further research in ML and its integration with other technologies to benefit the healthcare sector. Some areas are indicated below:

·        Great scope to develop strong Decision Support (DSS) system to help doctors and administrators of healthcare services.

·        Optimizing ML capability in managing medical claims, detecting fraud, improving clinical workflows, and predicting hospital-acquired infections.

·        Research for improving response time and ensuring better patient satisfaction.

·        There's a big scope for using NLP methods to study patients’ emotions and behaviour.

·        Multi-Media Analytic. Integration of patient information (text, image, audio, video, graphics) coming through multimedia and the Internet.

·        DNA sequencing. ML can help to apply super imaging technique and Natural Language Processing (NLP) to classify DNA sequencing.

·        Cancer Detection. ML can help to detect cancer based on the cell information provided by the Support Vector Machine (SVM)

·        Drug Discovery. ML can help Support Vector Regression (SVR) for disease-drug diagnostics and drug discovery.

·        Detect Mental Disorders. Use ML algorithms for early detection of mental disorders and suggested treatment.

·         Detecting Diabetes. ML can assist to analyze data like blood pressure, heart rate, and cholesterol level tests using Deep  Neural Networks (DNN) for detecting diabetes.

·        Cloud Computing and Big Data Technologies. Multimedia information comes like a flood and needs cloud computing and Big Data support. These need integration with ML.

·        5G Network. It is expected that 5G mobile networks will be available by mid-2021 to replace existing 4G networks. This will pave the way for ML drawing synergy from IoT / IoMT.

·        Cyber Security. During 2018-2019, some terrorist organizations/hackers had succeeded to disrupt operations of hospitals/banks/airports in many countries. Healthcare organizations/hospitals and drug development labs are equally vulnerable and would require foolproof cybersecurity.

·        Patient data privacy. This is a sensitive issue since multimedia users and service providers could misuse patient health data. There has to be a strong policy to protect a patient’s health data.

ML Jobs Opportunities. Today, a career in AI /ML field is most lucrative and highly paid for which both new and experienced professionals of IT/ Computing software and Mathematics/ Statistics fields are looking for an opportunity to join the AI/ML related jobs.  All such professionals who have an analytical mind and creative thinking can have career transition into ML area. However, the ML job is not simple software programming but is a very challenging and highly specialized field. It requires a judicious mix of master/graduate-level knowledge in software programming and statistics/mathematics. An entry-level ML professional requires an undergraduate degree in IT/ computer science with good knowledge of Numerical Methods and statistics.

A recent market survey conducted in the USA during mid, 2019 has revealed that ML market could grow to $8.81 billion by 2022. These estimates will be modified post-COVID 19. Jobs related to ML in various other sectors have been briefly covered in earlier chapters. In the healthcare sector, drug manufacturing companies, drug research labs, referral /research hospitals/institutes and healthcare organizations are recruiting professionals with experience in AI/ML. There is ever increasing demand for  ML Engineers, Data Scientists and  Product Analysts. Keeping in view large demand of AI / ML professionals, many technology training companies /institutes, colleges, universities and IIITs/ IITs are offering specialised training in ML.

Skill sets required for ML professionals. Skills in Python, “R”, MATLAB, Big Data and Cloud Computing is essential. Knowledge of Developing algorithms and predictive models are required for ML related jobs. Professionals seeking ML jobs in healthcare sectors would also require expertise in NLP, Neural Networks (NN), Linear Regression Analysis and Statistical Analysis, Image processing and computerised radiotherapy. Some skills common  to AI / ML jobs are briefly given below:

·     Programming language. Python and R. 

·     Computing Hardware. Knowledge of high-end computing machine, Operating System and GUI having powerful Numeric processor, very high-speed Graphics Processor Unit (GPU).

·     Software Tools. Experience in Cloud Computing, Amazon (AWS), Azure Microsoft or Google Cloud or IBM Cloud.  

·       Big Data and Analytic. Health Analytics, Hadoop, HIVE. PIG and Mongo DB are essential tools.

·     Natural Language Processing (NLP): NLP. helps to train a computing machine to understand both written text as well as voice. NLP techniques are needed to extract intelligence from unstructured text from documents and multi-media.

·     Machine Reasoning (MR). MR allows the ML system to make inferences based on data and also fill in the blanks for incomplete data.

·     Mass Open-source Online Courses (MOOC) Free online. MOOC courses in ML are available through  Udversity, Coursera, MIT, Stanford, Barkley, IIITs/IITs (India) and many other leading government / private universities, particularly those offering Liberal Arts programs. One can browse on Google to find a suitable and affordable program for full course with certification. Some basic online courses are free while certification may cost from Rs 750 to Rs 2500 of 6 to 12 months duration. Ministry of Human Resource and Development (MHRD), India has started a free MOOC online course in AI. This course is available on the UGC MOOC portal, a vertical of the SWAYAM initiative of MHRD. This course has 36 learning modules. One could choose as per his/her professional requirement.

·      Practice on live projects. After you get a good grasp of various ML algorithms and programming languages, you must undertake small projects by participating in online data-science hackathons /competitions. Healthcare employer would look for hand-on experience of using ML on real-life healthcare projects. The real expertise for getting good ML job comes through extensive practice on ML projects.

Indian Training Institutes /Universities/IITs. Keeping in view the global demand for AI / ML qualified professionals, many technologies training companies /institutes in cosmopolitan cities are offering specialised training. These courses are online / in-campus full time/ part-time for 6-18 months. One can customise training modules to meet a specific job requirement. Likewise, IISc Bengaluru, five top IITs, (Bombay, Madras, Delhi, Kanpur and Kharagpur) out of 23 IITs and two out of 23 IIITs (Hyderabad, Bengaluru), five top NITs out of 31 NITs and some select universities in India, have started offering postgraduate degree/ diploma in ML. Their training program is based on  Cloud Computing of (Microsoft, IBM, Google, Amazon), Data Science, Statistics, Applied Mathematics, Neural Network, Python. Big Data and Hadoop. They provide adequate hands-on training in modelling, coding programming data accessing and testing /validating. Likewise, AIIMS New Delhi is offering PG diploma in ML and IIM Bengaluru and NIMHANS, Bengaluru are jointly offering a PG diploma in ML. One must remember that it is not an easy choice to join the ML course and hope to gain expertise. One must do a thorough search on Google and decide on feedback and cost.

IT Majors offering software tools and training programs. IBM, Microsoft, Google and Amazon have taken big initiatives for developing training programs for supporting ML learning and practising. Amazon SageMaker and IBM,s Watson provides software tools for every data scientist to quickly learn, build, and deploy ML models. Likewise, Microsoft and Google have developed software tools, which are easy to learn, have fast take-off and facilitate in becoming ML professional. While registering and participating in such fast track programs, one must be regular in attending classes, submitting assignments and pay a fee for taking the test and get certified.

Private training institutes offering ML training and Jobs. At present, ML training institutes are located mainly in cosmopolitan cities like Mumbai, Pune, Bengaluru, Chennai, Hyderabad, Bhubaneswar, Ahmedabad, Delhi NCR (Gurgaon / Noida). They train fresh graduates, as well as employ, experienced professionals as their faculty to teach ML skills to young professionals. You could just Google it and find the right institute and its mode of teaching, project work, duration and cost. Some of these skills are common to AI, ML, Robotic and Gaming applications. Some technology leaders and consulting companies like Accenture Solutions, DEL international, Microsoft are also recruiting and training ML professionals Hence there are plenty of jobs in India and abroad for manpower skilled in AI/ML.  

Adapting ML.  It is a big challenge to create an ML model, evaluate its suitability, performance and efficiency and make its outcomes acceptable to the hospital/laboratory physicians. Rapid advances in computing technology, handling of unstructured data and increased use of Robotics, 5G, Multi-Media (MM) devices have enhanced productivity and quality of ML linked applications. To deliver accurate and quick results, ML algorithms must be continuously refined based on the outcome which reflects the current status of patient/disease.  ML relies on efficient OOP  framework like Python and “R”, and to some extent on MATLAB. Therefore, you must learn such new technologies faster and be proficient. If you implement the ML model successfully, it can pre-process a lot of healthcare data and predict the outcome for medical experts to speed up their decision-making. For successful implementation of ML in the healthcare sector, it is most essential to keep various algorithms, hardware, application software and data interfaces fully updated/upgraded.  From playing a critical role in patient care, billing, and medical records; ML technology is helping to develop alternate staffing models for providing better healthcare at reducing administrative costs.

Summary. Rapidly advancing technologies necessitate us to learn and share emerging technologies like ML, Multimedia, 5G and Cloud Computing to remain current and relevant. ML has a great role in optimising statistical modelling, disease-drug diagnostics and drug discovery. The tremendous speed of information retrieval and parallel processing of both structured and unstructured health-data and availability of GPUs, ML can help mining bio-information and ensure early detection of the virus. Indeed. ML brings both challenges and opportunities to the healthcare sector. The availability of  Cloud Computing, Big Data analytic, Open-Source Software ( OSS) like Python, R, TensorFlow and scikit-learn will provide much-needed impetus for widespread acceptance of using ML in the healthcare sector.

ML is invaluable for the Research & Development Work in the complex fields of Bio-Medical Engineering and Diagnostics. ML should not be treated as a black-box but used intelligently to supplement human efforts in providing good, timely and affordable healthcare. If things go wrong, you cannot sue the machine but the team who developed the algorithm. It is well known that clinical trials are costly and very time consuming and some of the trials may take two years or more to complete. ML has great potential to accelerate clinical trials and research work. Applying ML-based predictive analytics can help researchers to draw inference from a wide variety of data. Most researchers feel that ML techniques will be able to analyze current data of individual patients, and data uploaded using multimedia devices from remote locations. This can warn physicians at the main hospital about likely emergency in lower-level hospitals/. PHCs

References.

1.     Abhishek Singh & Karthik Rama Subramanian, Machine Learning using R, with time series and industry-based uses in R. 2019

2.     Arjun Panesar. Machine Learning and AI for Healthcare: Big data for improved health outcomes, 2019

3.     Jared Dean, Big data, Data mining and Machine Learning value creation for business leaders and practitioners, 2014

4.     Sarbjit Singh, Career Challenges during Global Uncertainty, PP 40, 47, chapter 5, NP Press, 2018