Detailed explanation of five major directions and 15 applications of big data medicine

    The medical industry has long faced the challenge of massive data and unstructured data. In recent years, many countries have actively promoted the development of medical information technology, which has enabled many medical institutions to have funds to conduct big data analysis. Therefore, the medical industry will be the first to enter the big data era along with the banking, telecommunications, insurance and other industries.

    In 1989, Gartner proposed the concept of BI. In 2008, Gartner further upgraded the concept of BI to Advanced Analytics. In 2011, McKinsey explained the concept of big data. Although the names are different, the problems they are trying to solve have never changed. It's just that today's big data analysis technology can handle more massive, diverse, and real-time (3V) data, that is, big data, than 20 years ago. Compared with BI 20 years ago, today's big data analysis can generate greater business value, and the development of big data storage and analysis technology has also benefited from the surge in data volume and the diversification of data types in business scenarios.

    Therefore, before implementing a big data analysis project, companies should not only know what technology to use, but also when and where to use it. In addition to Internet companies that have started to use big data earlier, the medical industry may be one of the traditional industries that first promoted big data analysis. The medical industry has long encountered the challenges of massive data and unstructured data, and in recent years many countries have been actively promoting the development of medical informatization, which has enabled many medical institutions to have funds to do big data analysis. Therefore, the medical industry will first enter the big data era together with industries such as banking, telecommunications, and insurance. McKinsey pointed out in its report that by eliminating institutional barriers, big data analysis can help the US medical service industry create $300 billion in added value a year. This article lists 15 applications in five major areas of the medical service industry (clinical business, payment/pricing, research and development, new business models, and public health). In these scenarios, the analysis and application of big data will play a huge role in improving medical efficiency and medical results.

    Clinical operation

    In terms of clinical operations, there are five major scenarios for big data applications. McKinsey estimates that if these applications are fully adopted, national healthcare spending in the United States alone will be reduced by $16.5 billion a year.

    1. Comparative effect study

    By comprehensively analyzing patient characteristics data and efficacy data and then comparing the effectiveness of multiple interventions, the best treatment path for a specific patient can be found.

    Efficacy-based research includes comparative effectiveness research (CER). Studies have shown that for the same patient, different medical service providers have different medical care methods and effects, and there are also great differences in costs. Accurate analysis of large data sets including patient vital signs data, cost data and efficacy data can help doctors determine the most clinically effective and cost-effective treatment methods. The implementation of CER in the medical care system will likely reduce overtreatment (such as avoiding treatments where side effects are more significant than efficacy) and undertreatment. In the long run, both overtreatment and undertreatment will have a negative impact on the patient's body and incur higher medical costs.

    Many medical institutions around the world (such as NICE in the UK, IQWIG in Germany, and the Canadian General Medicines Inspection Agency) have started CER projects and achieved initial success. The Recovery and Reinvestment Act passed by the United States in 2009 was the first step in this direction. Under this act, the Federal Coordinating Committee for Comparative Effectiveness Research was established to coordinate comparative effectiveness research across the federal government and allocate $400 million in funding. If this investment is to be successful, there are still many potential problems to be solved, such as the consistency of clinical data and insurance data. At present, in the absence of EHR (electronic health record) standards and interoperability, the large-scale and hasty deployment of EHR may make it difficult to integrate different data sets. Another example is the issue of patient privacy. It is not easy to provide enough detailed data to ensure the validity of the analysis results while protecting patient privacy. There are also some institutional problems. For example, current US law prohibits Medicare and Medicaid Services (medical service payers) from using cost/benefit ratios to make reimbursement decisions, so even if they find better methods through big data analysis, it is difficult to implement.

    2. Clinical decision support system

    Clinical decision support systems can improve work efficiency and the quality of diagnosis and treatment. Current clinical decision support systems analyze the entries entered by doctors and compare them with the differences in medical guidelines, thereby reminding doctors to prevent potential errors, such as adverse drug reactions. By deploying these systems, healthcare providers can reduce the rate of medical accidents and the number of claims, especially those caused by clinical errors. In a study of the Metropolitan Pediatric Intensive Care Unit in the United States, the clinical decision support system reduced the number of adverse drug reaction events by 40% within two months.

    Big data analysis technology will make clinical decision support systems smarter, thanks to the increasing ability to analyze unstructured data. For example, image analysis and recognition technology can be used to identify medical imaging (X-ray, CT, MRI) data, or to mine medical literature data to build a medical expert database (like IBM Watson does), so as to provide doctors with diagnosis and treatment recommendations. In addition, clinical decision support systems can also allow most of the workflow in the medical process to flow to nursing staff and assistant doctors, freeing doctors from simple consultation work that takes too long, thereby improving treatment efficiency.

    3. Medical data transparency

    Improving the transparency of medical process data can make the performance of medical practitioners and medical institutions more transparent, and indirectly promote the improvement of medical service quality.

    Based on the operational and performance data sets set by healthcare providers, data analysis can be performed and visual flowcharts and dashboards can be created to promote information transparency. The goal of the flowchart is to identify and analyze the sources of clinical variation and medical waste, and then optimize the process. Simply publishing cost, quality, and performance data, even without corresponding material rewards, can often promote performance improvements, enabling healthcare service organizations to provide better services and become more competitive.

    Data analysis can streamline business processes, reduce costs through lean production, and find more efficient employees who meet the needs, thereby improving the quality of care and bringing a better experience to patients, and also bringing additional performance growth potential to medical service organizations. The Centers for Medicare and Medicaid Services in the United States is testing dashboards as part of building an active, transparent, open, and collaborative government. In the same spirit, the Centers for Disease Control and Prevention in the United States has publicly released medical data, including business data.

    Publicly releasing quality and performance data can also help patients make more informed health care decisions, which will also help healthcare providers improve their overall performance and become more competitive.

    4. Remote patient monitoring

    Collect data from remote monitoring systems for chronic patients and feed analysis results back to monitoring devices (to see if patients are following doctor's orders) to determine future medication and treatment plans.

    In 2010, there were 150 million patients with chronic diseases in the United States, such as diabetes, congestive heart failure, and hypertension, and their medical expenses accounted for 80% of the medical costs of the health care system. Remote patient monitoring systems are very useful for treating patients with chronic diseases. Remote patient monitoring systems include home heart monitoring devices, blood glucose meters, and even chip tablets. After being ingested by patients, chip tablets transmit data to the electronic medical record database in real time. For example, remote monitoring can remind doctors to take timely treatment measures for patients with congestive heart failure to prevent emergencies, because one of the signs of congestive heart failure is weight gain due to water retention, which can be prevented through remote monitoring. More benefits are that through the analysis of data generated by remote monitoring systems, patients' hospital stays can be reduced, the number of emergency visits can be reduced, and the goal of increasing the proportion of home care and the number of outpatient doctor appointments can be achieved.

    5. Advanced analysis of patient records

    Applying advanced analytics to patient records can determine who is more susceptible to certain diseases. For example, applying advanced analytics can help identify patients who are at high risk for diabetes so they can receive preventive care programs as early as possible. These methods can also help patients find the best treatment options from existing disease management programs.

    Payment/Pricing

    For medical payers, big data analysis can better price medical services. Taking the United States as an example, this will have the potential to create $50 billion in value each year, half of which comes from reducing national medical expenses.

    1. Automation system

    Automated systems (e.g., machine learning techniques) detect fraud. Industry insiders estimate that 2% to 4% of medical claims are fraudulent or unreasonable each year, so detecting claims fraud has great economic significance. Through a comprehensive and consistent claims database and corresponding algorithms, claims accuracy can be detected and fraud can be detected. This fraud detection can be retrospective or real-time. In real-time detection, automated systems can identify fraud before payment occurs, avoiding significant losses.

    2. Pricing plans based on health economics and efficacy research

    In terms of drug pricing, pharmaceutical companies can participate in sharing treatment risks, such as formulating pricing strategies based on treatment effects. The benefits to medical payers are obvious, which is conducive to controlling healthcare cost expenditures. For patients, the benefits are more direct. They can get innovative drugs at reasonable prices, and these drugs have been studied based on efficacy. For pharmaceutical product companies, better pricing strategies are also beneficial. They can obtain a higher probability of market access, and can also obtain higher revenues through innovative pricing schemes and more targeted launch of effective drugs.

    In Europe, there are now some pilot projects on drug pricing based on health economics and efficacy.

    Some healthcare payers are using data analytics to measure the services of healthcare providers and set prices based on the level of service provided. Healthcare payers can pay based on medical outcomes, and they can negotiate with healthcare providers to see if the services provided by healthcare providers meet certain benchmarks.

    Research and Development

    Medical product companies can use big data to improve R&D efficiency. Taking the United States as an example, this will create a value of more than $100 billion per year.

    1. Predictive Modeling

    During the research and development phase of new drugs, pharmaceutical companies can use data modeling and analysis to determine the most efficient input-output ratio, thereby allocating the best resource combination. The model is based on data sets before the drug clinical trial phase and early clinical phase data sets to predict clinical results as promptly as possible. Evaluation factors include product safety, effectiveness, potential side effects, and overall test results. Predictive modeling can reduce the research and development costs of pharmaceutical product companies. After predicting the clinical results of drugs through data modeling and analysis, research on suboptimal drugs can be postponed, or expensive clinical trials on suboptimal drugs can be stopped.

    In addition to R&D costs, pharmaceutical companies can also get returns faster. Through data modeling and analysis, pharmaceutical companies can bring drugs to market faster, produce more targeted drugs, and drugs with higher potential market returns and treatment success rates. It used to take about 13 years for a new drug to be developed and brought to market. The use of predictive models can help pharmaceutical companies bring new drugs to market 3 to 5 years earlier.

    2. Statistical tools and algorithms to improve clinical trial design

    The use of statistical tools and algorithms can improve the level of clinical trial design and make it easier to recruit patients during the clinical trial stage. By mining patient data and evaluating whether the recruited patients meet the trial conditions, the clinical trial process can be accelerated, more effective clinical trial design suggestions can be made, and the most suitable clinical trial base can be found. For example, trial bases with a large number of potentially eligible clinical trial patients may be more ideal, or a balance can be found between the size and characteristics of the trial patient population.

    3. Analysis of clinical trial data

    Analyzing clinical trial data and patient records can identify more indications for drugs and discover side effects. After analyzing clinical trial data and patient records, drugs can be repositioned or marketed for other indications. Real-time or near-real-time collection of adverse reaction reports can promote pharmacovigilance (pharmacovigilance is a safety assurance system for marketed drugs that monitors, evaluates and prevents adverse drug reactions). Or in some cases, clinical trials suggest some situations but there is not enough statistical data to prove them. Now analysis based on clinical trial big data can provide evidence.

    These analysis projects are very important. We can see that the number of drug withdrawals has hit new highs in recent years, and drug withdrawals can be devastating to pharmaceutical companies. The withdrawal of the painkiller Vioxx from the market in 2004 caused Merck to lose $7 billion, resulting in a 33% loss in shareholder value in just a few days.

    4. Personalized treatment

    Another promising big data innovation in the field of R&D is the development of personalized treatments through the analysis of large data sets (such as genomic data). This application examines the relationship between genetic variation, susceptibility to specific diseases and response to specific drugs, and then considers individual genetic variation factors in the drug development and medication process.

    Personalized medicine can improve health care outcomes, such as providing early detection and diagnosis before patients develop symptoms of disease. In many cases, patients respond differently to the same treatment, in part because of genetic variation. Using different treatments for different patients, or adjusting drug dosages based on the patient's actual situation, can reduce side effects.

    Personalized medicine is still in its early stages. McKinsey estimates that in some cases, it can reduce medical costs by 30% to 70% by reducing the amount of prescribed drugs. For example, early detection and treatment can significantly reduce the burden of lung cancer on the health system because the cost of early surgery is half that of later treatment.

    5. Analysis of disease patterns

    By analyzing disease patterns and trends, medical product companies can make strategic R&D investment decisions and optimize their R&D priorities and resource allocation.

    New business models

    Big data analysis can bring new business models to the medical services industry.

    1. Aggregate patients’ clinical records and medical insurance datasets

    Aggregating patients' clinical records and medical insurance data sets and performing advanced analysis will improve the decision-making capabilities of medical payers, medical service providers, and pharmaceutical companies. For example, for pharmaceutical companies, they can not only produce drugs with better efficacy, but also ensure that the drugs are marketable. The market for clinical records and medical insurance data sets has just begun to develop, and the speed of expansion will depend on the speed at which the healthcare industry completes the development of EMR and evidence-based medicine.

    2. Online platforms and communities

    Another potential business model for big data startups is online platforms and big data, which have already generated a lot of valuable data. For example, PatientsLikeMe.com, a website where patients can share their treatment experiences; Sermo.com, a website where doctors can share medical insights; Participatorymedicine.org, a website run by a nonprofit organization to encourage patients to actively participate in treatment. These platforms can become a valuable source of data. For example, Sermo.com charges pharmaceutical companies to allow them to access member information and online interaction information.

    Public Health

    The use of big data can improve public health monitoring. Public health departments can quickly detect infectious diseases, conduct comprehensive epidemic monitoring through a nationwide patient electronic medical record database, and respond quickly through integrated disease monitoring and response programs. This will bring many benefits, including reduced medical claims expenditures, lower infectious disease infection rates, and health departments can detect new infectious diseases and epidemics more quickly. By providing accurate and timely public health advice, public health risk awareness will be greatly improved, while the risk of infectious disease infection will also be reduced. All of this will help people create a better life.