Administrative costs substitute more than $300 billion annually in the US healthcare system budget. It is 15 percent of all healthcare expenditures.
Accenture states that AI applications potentially can generate $150 billion in annual savings. In healthcare, Artificial Intelligence implies technologies that enable machines to sense, comprehend, act and learn to perform administrative and clinical healthcare functions.
Frost&Sullivan predicts the healthcare AI market to get to $6.6 billion by 2021, a 40% growth rate. It also says that there is a potential in AI to improve outcomes by 20-4- percent and cut costs by 50 percent.
AI has become extremely popular in the healthcare industry: it is expected to grow to $6.6 billion by 2021 with a compound annual growth rate of 40 percent. CB Insights predicts the health AI market to grow 10 times.
Healthcare IT News and HIMSS report that 35 percent of healthcare organizations are planning to adopt artificial intelligence within 2 years, and more than half — within 5. The top spheres to apply AI in healthcare are population health, clinical decision support, patient diagnosis and precision medicine. Other spheres of interest are hospital and physician workflow, security, revenue cycle and drug discovery.
To the top 10 AI applications, belong robot-assisted surgery, virtual nursing assistants, administrative workflow assistance, fraud detections, dosage error reduction, connected machines, clinical trial participant identifier, preliminary diagnosis, automated image diagnosis, cybersecurity.
Cognitive robotics integrates pre-operational medical records with real-time operational metrics available. This enhances physician’s precision incorporating data from the actual surgical experiences. It reduces the length of patient’s stay by 21 percent.
Remote patient monitoring
Remote patient monitoring is a great feature. And when it delivers notifications and alerts to the medical personnel only when a patient needs a treatment, it gets even better, reducing unnecessary hospital visits.
Voice-to-text transcription, an administrative workflow assistant, helps to concentrate on the treatment process but not non-patient activities. It’s been estimated to save 17 percent of doctors time and 51 percent for registered nurses.
ZocDoc, medical appointment booking app, has acquired a new feature: Insurance Checker powered by artificial intelligence. It deciphers and verifies the health insurance. CAQH report states that healthcare providers spend a quarter of their time dealing with insurance.
Another large sector that consumes a lot of time and fundings is customer support. AI-infused automated chat-bots can substantially decrease the costs. Juniper Report forecasts that organizations can annually save $8 billion worldwide by 2022. Chatbots can also save up to
4 minutes per inquiry that equals to around $0.60 per interaction.
Chatbots with AI use such technologies as natural language processing, knowledge management and sentiment analysis. Natural language processing is used to understand what a user is asking about, their intent and the level of confidence. The technological methodology supports conversational flow and responses.
Knowledge management systems document common questions, answers and solutions that are accumulated throughout the product lifetime. Sentiment analysis systematically identifies, extracts, quantifies and studies affective states and subjective information. Simply put, it aims at defining the user’s attitude or emotional reaction.
Virtual assistance is another sphere AI can be applied to. Dragon Medical Virtual Assistant is developed to manage clinical workflow. It covers voice recognition and biometrics and tech-to-speech. It can be also integrated with EHR. It acts like Siri when a doctor can just ask “Show me the latest results”, and the assistant returns the up-to-date information.
AI can help with lab examinations, as well. The doc.ai app is planned to interpret the lab results. It will use natural language processing to converse with patients via an app.
AI can even be used for the security purposes. In line with the recent malware, this is a topical sphere. Aetna allows its users to apply one of their biometric factors as one of their authentication attributes. Later on, the risk engine analyzes not only the device attributes but also user behavior, like the way a user holds his or her phone and matches these attributes with previous behavior model.
Huge EHR systems, like Epic, are planning to AI for computer-assisted physician documentation tool into the modules for clinical documentation improvements. It will help doctors at the point-of-care to better understand reimbursement and risk-adjustment factors. With deep learning and natural language processing, the system can determine clinical info in electronic medical records and notify a doctor about missing data or required clarifications.
IT giants are getting actively involved in healthcare AI. For instance, Google’s deep learning algorithm can detect diabetic retinopathy with more than 90 percent accuracy. It’s been trained on a large scope of fundus images. They are also working on detection of cancer spreading to adjacent lymph nodes.
Apple has bought AI company Lattice that uses AI to turn unstructured “dark data” generated by digital interactions into structured one. With its Healthcare NExT, Microsoft combines AI with the cloud, research and industry partnerships to improve healthcare.
Alphabet-owned Verily together with Stanford Medicine and Duke University School of Medicine has launched the Project Baseline Study to accumulate phenotypic health data to create a baseline map of human health to get insights into health disease transitions. The information includes clinical, imaging, self-reported, behavioral, and that from sensors and biospecimen samples of 10,000 participants.
The most barriers to AI adoption is the feeling that the technology is still way too fresh with not enough successful business cases. Other obstacles are regulatory barriers, interoperability with legacy hospital IT systems, and serious limitations on access to crucial medical data needed to build powerful health-focused algorithms in the first place. It has been estimated to take around 2 years of work on infrastructure to progress with the AI. SaaS platforms are seen as an opportunity to save time and costs on this.
Budget has appeared to be another barrier on the way of AI in healthcare. 57 percent of respondents in the Intel report defined costs as the main challenge in the AI adoption. Usually, it is hard for the IT directors to show the return on investment rate with AI. The senior executives want to see the results that are not immediately seen with the new technologies.
68 percent of respondents chose the lack of technical expertise and human capital to the top 3 main obstacles. Data scientists are the top-wanted employees. To the possible solutions belong relocation, training and outsourcing.