Employing AI to Improve Healthcare Delivery | ORF

Artificial Intelligence (AI) has already made inroads into healthcare. The growing availability of healthcare knowledge has allowed for the deployment of highly effective AI methods to help drug growth in addition to medical observe. These run the gamut from machine studying (ML) methods that analyse structured knowledge, to pure language processing (NLP) that may extract significant info from unstructured knowledge, corresponding to medical doctors’ notes. This temporary examines the guarantees, dangers and challenges of adopting AI in healthcare, drawing on proof from throughout the globe. It explores how AI can reshape the Indian healthcare ecosystem and what which means for present stakeholders together with hospitals and medical doctors, in addition to know-how and insurance coverage firms.


Attribution: Ashish Airon and Ayesha Jhunjhunwala, “Employing A.I. to Improve Healthcare Delivery,” ORF Issue Brief No. 385, July 2020, Observer Research Foundation.


Introduction

To perceive the long-term impression of AI on healthcare, it’s essential to first perceive its affect within the short- to medium-term. The growing use of AI within the business may have necessary ramifications on the healthcare system. In the foreseeable future, AI will grow to be integral to not solely acute care delivered by means of hospitals and clinics but in addition a lot of what’s thought-about wellness, corresponding to diet, train and sleep. It will precipitate a shift from a largely reactive system constructed to answer harm and sickness, to at least one geared in the direction of early detection and prevention. These modifications, in flip, will result in shifts in key stakeholders and their roles: insurance coverage firms are prone to play a a lot bigger function in figuring out care protocols and driving early detection; and pharmaceutical firms will more and more uncover and develop medicine in partnership with know-how firms. Companies corresponding to Apple and Google are already investing closely in healthcare, partnering with hospitals and pharmaceutical firms to develop AI-based instruments that scale back price, improve entry and enhance high quality.

In India, the adoption of AI in healthcare remains to be in its nascent stage. While varied homegrown know-how firms have raised funding, e.g. Qure, Niramai and Sigtuple, deployment at scale stays a problem, attributable to low ranges of digitisation, stakeholder resistance attributable to low consciousness, and a tough regulatory surroundings. 

The Use of AI in Healthcare

The present mannequin of healthcare supply is a operate of obtainable info. When a affected person experiences signs, they bear diagnostic assessments, usually underneath the steerage of a health care provider—both a basic doctor or a specialist. Pathologists use machines to detect blood counts and abnormalities, and radiologists learn X-rays and report their findings. The attending physician then aggregates the data and gives a remedy plan.

A major problem that this workflow presents is that the information collected is often saved in varied remoted databases, usually in several codecs. Before AI might be deployed to enhance outcomes, it should combine this knowledge in a dependable and inexpensive method. The creation of a knowledge lake[1] is a typical first step in the direction of consolidating from a number of silos. An information lake helps consolidate uncooked knowledge right into a single supply to allow algorithms to make diagnoses or suggestions primarily based on full datasets. AI-based algorithms can then research knowledge utilization, together with entry frequency, generally used parts, and person roles, which permits them to effectively clear up complicated integration issues, e.g. integrating giant volumes of knowledge streamed by IoT units corresponding to wearables. AI instruments will enhance the standard of care in 4 important areas, with out introducing any important modifications within the course of from the attitude of the sufferers.

Detection

The use of AI will facilitate the early detection of ailments. Currently, remedy is sought solely after experiencing signs. However, knowledge from steady monitoring of well being can routinely set off interventions. There is now a burgeoning ecosystem of monitoring units, corresponding to smartwatches, phone-based sleep trackers, and wearable blood-pressure screens. For instance, the Apple Watch, which has been licensed by the FDA as a medical gadget, can passively perform electrocardiograms (ECGs) at common intervals and generate alerts upon recording a worrying pattern. Moreover, since early detection is vital to decreasing prices and bettering outcomes, insurance coverage firms, too, at the moment are pushing for normal testing and monitoring.

Diagnosis

In the long run, diagnoses will more and more be aided by AI, already evident in imaging-based modalities corresponding to radiology and pathology. Radiology has been main the cost, with firms specializing in constructing know-how that may carry out computer-aided detection (CADe) and computer-aided analysis (CADx). Specialised know-how firms, corresponding to Zebra in Israel, are utilizing huge imaging databases to coach algorithms that may detect abnormalities in X-Ray, MRI and CT Images. Companies that make imaging {hardware}, e.g. Siemens, have begun providing supporting AI software program such because the AI-Rad Companion, which might be built-in into workflows to scale back the burden of repetitive duties and improve diagnostic precision. The AI-Rad Companion at the moment encompasses a collection of choices together with particular modules for chest CTs, mind MRIs and prostate MRIs. Arterys provides FDA-approved software program to assist scale back reporting time by 30 % for cardiac and chest X-rays. Algorithms have additionally achieved dermatologist-level accuracy within the classification of pores and skin most cancers.[2] QuantX and Profound AI have constructed FDA-approved software program that helps radiologists within the evaluation and characterisation of breast abnormalities in MRI photographs. Pathology just isn’t far behind. For occasion, New York-based PathAI is growing machine-learning know-how to assist pathologists scale back errors in most cancers diagnoses and develop strategies for individualised remedy. Their collaborators embody Bristol Myers-Squibb in addition to the Bill and Melinda Gates Foundation.

Treatment

AI will present ‘decision support’ for protocol-based remedy of routine circumstances. It is particularly helpful in conditions that contain substantial knowledge, whereby traits could also be hidden by ‘noise’ that obscures the precise downside. One such instance is in sepsis detection and remedy. Sepsis is the main reason behind loss of life in American hospitals, and Johns Hopkins estimates that at the very least 80 % of those deaths are preventable if identified in time.[3] AI-based analysis, primarily based on vitals and lab knowledge, can assist predict the danger of sepsis effectively earlier than it turns into obvious to clinicians. Since the remedy for sepsis is often primarily based on clearly outlined algorithms, AI-based instruments may even be used to determine the simplest course of remedy and drug dosages primarily based on a affected person’s historical past and profile. Companies corresponding to Google and Bayesian Health are already engaged on AI-driven early detection, in partnership with hospitals corresponding to Duke University Hospital and Sentara.[4] Additionally, AI-based instruments have efficiently helped present look after persistent ailments corresponding to diabetes[5],[6] and Chronic Obstructive Pulmonary Disorder (COPD).[7]

Personalised Medication

Personalised remedy, or precision drugs, leverages a person’s distinctive genetic knowledge, and is an bold and rising area of medication. Large multidimensional organic datasets are used to optimise analysis, prognosis and remedy. Interventions are tailor-made primarily based on a affected person’s genetics, surroundings and biology. Such a way has already confirmed helpful in predicting the danger of cancers[8],[9] and cardiovascular ailments.[10],[11] Pharmacogenomics, the research of how genes have an effect on drug response, mixed with AI-based instruments are already getting used to efficiently predict remedy outcomes.[12],[13],[14] This helps clinicians not solely selected between a number of remedy choices but in addition forestall hostile drug reactions. However, one of many main challenges in deploying precision drugs at scale is the supply of intensive and high-quality datasets, which is essential to this course of.

The Impact of AI in Healthcare

Healthcare Delivery

The use of AI in several areas of healthcare will cumulatively drive a shift in each the focus and the format of healthcare supply. The focus will shift from reactive interventions and acute care to pre-emptive care and common monitoring, making use of equally to chronically in poor health, high-risk, and wholesome people. Consequently, a lot of what’s at the moment thought-about “wellness” will ultimately grow to be a key part of healthcare supply.

The format of care supply may also change, shifting partially from brick-and-mortar amenities corresponding to hospitals to step-down clinics and even sufferers’ properties. Part of this shift can be pushed by the expansion and proliferation of distant monitoring units. The growth of the units and the telecommunications bandwidth to transmit giant volumes of real-time knowledge usually are not depending on AI. However, AI is required to course of the big volumes of knowledge, determine warning alerts primarily based on 1000’s of knowledge factors, and alert the related caregivers in time. A key driver of this variation would be the progress of AI-augmented telemedicine, which has the potential to deal with urgent wants in two methods.

AI could make a helpful contribution in areas the place it’s tough to co-locate sufferers with certified physicians. Companies corresponding to Buoy, Babylon and Lemonaid have already deployed AI-augmented know-how to assist triage, diagnose and prioritise sufferers. They have developed chatbots that may work together with sufferers to assist present a analysis for instances the place the signs correspond to widespread issues corresponding to urinary tract infections. More sophisticated instances are referred to specialists. Not solely does this scale back the burden on frontline care workers, corresponding to GPs and ER medical doctors, additionally it is helpful for populations that shouldn’t have bodily entry to certified clinicians. Babylon not too long ago entered right into a 10-year partnership with the Rwandan authorities[15] to assist it handle the nation’s acute scarcity of medical doctors.

AI-augmented telemedicine additionally provides worth in conditions the place giant volumes of knowledge should be captured and studied to offer care. Thus, a rising use case for AI-augmented telemedicine is in distant monitoring, whereby knowledge is tracked by sensors corresponding to EEGs, ECGs, blood glucose screens, heart-rate screens, pulse oxygen and respiratory screens. This ecosystem can present a security web for sufferers with persistent circumstances corresponding to COPD, diabetes and coronary heart illness. It can even assist in decreasing the size of a affected person’s hospital keep by permitting ward-level monitoring to be efficiently carried out at house. These instruments are particularly highly effective when built-in with the broader healthcare ecosystem, together with tertiary care and specialist referrals, drugs supply, and emergency response companies. Companies corresponding to Ten3t in India, Current in Scotland, and Eko within the US have made important progress in bringing such options to the market.

Pharmaceutical Companies

AI has already begun demonstrating ends in the pharmaceutical business, particularly in drug discovery, drug growth and pharmacovigilance (PV). Drug discovery and growth are lengthy and costly processes, with many medicine usually taking upwards of 10 years and over US$ 1 billion to make it to the market. AI can handle each ache factors by decreasing the time required and the event price. It can also be being leveraged for PV, i.e. to reinforce drug security.

Drug discovery, regardless of enhancements in medical know-how, has grow to be slower and dearer over time. Pharmaceutical firms are underneath stress and more and more turning to partnerships with know-how firms that may assist them use AI to speed up the drug-discovery course of. Companies corresponding to Recursion Pharma try to deal with this by constructing drug-delivery platforms that reveal new therapeutic candidates. In 2016, Pfizer and IBM Watson introduced a partnership to speed up drug discovery in immuno-oncology. In 2017, Sanofi partnered with Exscientia to make use of their AI-driven platform to assist determine and design medicine that may deal with metabolic ailments.[16] Exscientia has additionally partnered with GSK for the same drug-discovery undertaking. These partnerships have already begun to yield outcomes: the GSK-Exscientia collaboration discovered its first candidate molecule in April 2019.[17]

Roughly half the time and price of getting medicine to market is spent on the medical trial part. These have excessive failure charges, attributable to prices related to figuring out and enrolling acceptable topics (86 % of trials don’t meet enrolment timelines),[18] monitoring them over the interval of the trial, managing non-adherence, and decreasing the drop-out price. Companies corresponding to Mendel.AI, Antidote and Deep 6 AI have developed platforms that can be utilized to effectively match sufferers with medical trials. Mayo Clinic not too long ago launched the outcomes of a research that discovered that utilizing IBM Watson helped improve trial enrolment by 80 %.[19] At the identical time, firms corresponding to Brite Health have focused a special ache level by growing options to assist handle affected person engagement over the course of a trial, which reduces the non-adherence and drop-out charges. Researchers have additionally efficiently used AI to design higher medical trials by iteratively adjusting drug dosages to optimise remedy.[20]

Major tech firms have been notably energetic on this area. The world’s second-largest pharmaceutical firm, Roche, is trialling Amazon’s Comprehend Medical instrument to assist match sufferers to out there medical trials.[21] Novartis and Microsoft have introduced a five-year partnership, the place the previous will draw on the latter’s AI experience to design a instrument for calculating private drug dosage to deal with macular degeneration, optimising the manufacturing of CAR-T cells, and shortening the time required to develop new medicine through the use of neural networks to generate, display screen and choose promising molecules.[22] Google, too, has collaborated with London-based DeepMind to construct AlphaFold, a mannequin that may assist predict protein construction, permitting scientists to focus on and design new cures extra effectively.

AI-based instruments are more and more being deployed to enhance drug security by monitoring manufacturing processes in actual time, figuring out efficacy, and accumulating and processing knowledge on hostile occasions (AEs). The mounting regulatory deal with security and the ensuing improve in compliance burden are necessary drivers for the rising use of AI, notably on the subject of AE case processing. Moreover, research point out that case processing consumes at the very least two-thirds of drug firms’ total PV funds.[23]  Pilots have already demonstrated the feasibility of making use of AI to automate security case processing and thereby favourably impression the strongest price driver of the general PV funds.[24] Similarly, AI instruments can be utilized in healthcare for post-marketing surveillance to check the results of medication on particular populations, particularly in mild of considerations that medical trials could not guarantee ample illustration. For occasion, Chazard et al. efficiently used knowledge from over 100,000 well being information to generate guidelines for the detection of AEs.[25] Thus, AI-based instruments have a two-fold utility on this space: retrospective AE detection in addition to potential AE prevention when embedded inside medical determination assist methods.

Stakeholders

AI-driven modifications will inevitably result in a shift in energy away from medical doctors and hospitals, as insurers and know-how firms grow to be more and more related gamers within the ecosystem. Companies that harvest and course of affected person knowledge, e.g. Apple and Fitbit, and those who perform genetic testing, will possess a helpful useful resource, which can be deployed in the direction of prevention, early detection, and higher analysis. Insurance firms are prone to utilise this knowledge to regulate premiums and drive prevention and early detection. The velocity of this shift will rely upon the speed of improve in insurance coverage penetration. For instance, it’ll speed up if the world strikes in the direction of an accountable care mannequin, the place insurers are paid a hard and fast quantity to take care of the well being of an outlined inhabitants, primarily based on goal final result indicators.

It is simply too early to foretell how healthcare will evolve in a rustic corresponding to India, the place insurance coverage penetration is low and nearly all of healthcare expenditure is personal. The authorities’s most up-to-date National Health Policy (2017) explicitly lists “progressively achieving Universal Health Coverage” as a purpose. To this finish, the doc outlines a purpose of accelerating well being expenditure to 2.5 % of the funds and over eight % of states’ budgets.[26] The said visions of schemes corresponding to Ayushman Bharat[27] and the National Health Stack proposed by NITI Aayog are additional indications that the federal government is eager to maneuver on this route. However, ground-level implementation stays patchy.[28]

Major know-how firms will grow to be more and more necessary stakeholders within the healthcare ecosystem. Already, the Google-Deep Mind partnership has efficiently developed algorithms to assist detect breast most cancers (the algorithm produced six % fewer incorrect diagnoses and 9 % fewer false negatives relative to common medical observe)[29] and diabetic retinopathy (utilizing a dataset from India and the US, their algorithm carried out marginally higher than a panel of ophthalmologists).[30] Google can also be partnering with different tech firms, corresponding to Viz.ai,[31] which constructed a instrument to analyse CT scans and notify healthcare professionals about potential strokes.

AI is unlikely to exchange healthcare professionals corresponding to medical doctors, nurses, diagnosticians and technicians and can solely function a “computer-aided” determination assist system. However, medical doctors will not be the one supply of reality. They will nonetheless be key decision-makers in each illness analysis and remedy (planning in addition to execution), however these duties will more and more grow to be a workforce effort, with AI-based machines facilitating higher analysis and remedy choices.

Employment within the Healthcare Sector

According to a 2019 report revealed within the Future Healthcare Journal, “To our knowledge thus far there have been no jobs eliminated by AI in health care.” Thus, any panic relating to AI creating mass redundancies in healthcare is untimely. However, not solely may this variation sooner or later however AI may additionally impression the workforce in different methods.

Moreover, AI may also change activity profiles throughout the healthcare workforce. It will permit clinicians to spend much less time gathering and monitoring knowledge and is prone to change a minority of routine duties throughout the workforce. A 2019 research[32] by the American Hospital Association discovered that AI may take over roughly one-third of the duties at the moment carried out by clinicians. The tasks of nurses and paramedical workers may shift away from work that may be automated and in the direction of computer-based work, corresponding to knowledge entry and reviewing routinely captured knowledge.

The elevated use of AI may also create redundancies in a small set of center and lower-end jobs, corresponding to affected person monitoring (usually performed manually by nurses at common intervals) and radiologists who report fundamental scans. Jobs which might be routine, repetitive, predictable, and the place the stakes are comparatively low can simply be automated. However, in the long term, research predict that AI will create at the very least as many roles as are destroyed,[33] e.g. roles in knowledge science, software program and gadget engineering, and knowledge governance. Further, given the worldwide scarcity of certified care professionals,[34] it’s unlikely that healthcare will witness important job losses within the foreseeable future.

Key Risks

Increasingly, healthcare is shifting in the direction of evidence-based and probabilistic automated fashions for decision-making, which comes with its personal set of dangers.

Biased Models

AI fashions are developed primarily based on datasets and “learn” utilizing the information supplied to them. However, datasets can have inherent biases, such because the underrepresentation of a sure race or gender. The use of biased datasets in treating or diagnosing marginalised group can inadvertently compromise outcomes and security. A 2016 report[35] that aggregated the outcomes from 2,511 research discovered that 81 % of individuals in genome-mapping research had been of European descent. Researchers who obtain these datasets to check ailments or develop AI-based fashions could unwittingly reproduce the identical biases of their work. According to a 2014 research,[36] the shortage of various analysis topics is likely one of the causes that black Americans are considerably extra prone to die than white Americans after a most cancers analysis.

Data Security

The assortment, storage and transmission of knowledge on a big scale drastically will increase the possibilities of inadvertent leaks or breaches of privateness. This, in flip, can result in secondary harms the place delicate well being info is misused, e.g. to discriminate in opposition to people when figuring out their eligibility for employment or housing. Such considerations have resulted within the adoption of standardised safety protocols for well being info, together with the Health Information Portability and Accountability Act, 1996 and the Fast Healthcare Interoperability Resources (FHIR) customary for the digital alternate of healthcare info. Moreover, the supply of private knowledge poses important dangers in areas with poor telecommunications or energy infrastructure. Therefore, counting on this knowledge for medical choices can delay care if the information turns into quickly or completely unavailable. This is particularly related for rural populations or for populations in growing and lower-income international locations, the place web entry just isn’t but dependable.

Inequity in Access

There is a ‘black box’ downside related to AI-based instruments, particularly with fashions developed utilizing neural networks. Unlike in conventional research, it’s obscure the connection between inputs and outputs in such fashions. While this isn’t a significant downside when the purpose is mere prediction, in analysis or remedy, the shortcoming to determine hyperlinks can pose a threat. If not successfully managed, these dangers cannot solely compromise affected person security and healthcare outcomes but in addition exacerbate present inequalities in entry to care. For instance, mannequin bias is prone to compromise look after segments of the inhabitants that have already got comparatively poor entry to care. Women usually tend to wait in emergency rooms and are much less prone to be given efficient painkillers;[37] they’re additionally extra prone to have physiological illnesses written off as psychological illnesses.[38] Models educated on present remedy protocols run the danger of “hard-coding” these inequalities and making them more durable to recognise, handle or reverse.

AI-based algorithms usually require important supporting infrastructure, corresponding to {hardware}, software program and telecommunications infrastructure. Thus, the deployment of AI in healthcare is prone to stay restricted to privileged areas, corresponding to prosperous cities, exacerbating present inequalities in entry to healthcare. Moreover, inequalities could prolong past socio-economic parameters. For occasion, though the US has enacted laws that prohibits employers and medical insurance firms from discriminating on the premise of genetic knowledge (the Genetic Information Nondiscrimination Act), this doesn’t apply to other forms of insurance coverage, corresponding to life insurance coverage.

Key Adoption Challenges

The potential for AI to scale back prices and enhance high quality in healthcare has already been established. However, large-scale adoption is a major problem and prone to stay so for the foreseeable future. This is especially as a result of constraints of legacy methods, workflows and cultural norms, that are tough to exchange. This is exacerbated by the shortage of appropriate supporting infrastructure, when it comes to not solely know-how but in addition regulatory infrastructure.

Lack of Data

The single greatest constraint on the event of AI-based instruments is the supply of high-quality datasets with sufficient knowledge to coach strong fashions. For instance, an imaging mannequin would require quite a few photographs, with the annotations on every required to be clear, bias-free, exhaustive, error-free, and introduced in a manner that the mannequin is able to processing. Moreover, the “feature space” being bigger than the observations (a standard case in healthcare attributable to restricted knowledge) ends in overfitting, whereby the mannequin doesn’t generalise effectively in manufacturing. Workflows are usually not set as much as produce huge portions of correctly annotated scans, and plenty of organisations are reluctant to share such scans for concern of affected person privateness being compromised. Once datasets are assembled and fashions developed, integrating these fashions into present methods and workflows is one other tough course of. For occasion, the AI-based diagnoses and proposals may not be out there as a part of the present software program ecosystem, e.g. radiology storage and reporting software program such because the RIS and PACS or the digital medical information which might be in use. The suggestion will then should be manually reviewed, which, in observe, is prone to constrain uptake in fast-paced and high-stress environments. Replacing present methods is an costly and dangerous endeavor. The promise of AI offering an incremental enchancment in care high quality usually doesn’t justify the prices or the dangers that will be incurred.

Resistance from the Medical Community

Another necessary constraint within the adoption of AI-enabled decision-support is resistance from the medical group. Doctors who’re unfamiliar with the know-how and don’t totally perceive the way it works are sometimes reluctant to make use of it. This is barely compounded by considerations of AI changing people and the absence of any clear medico-legal pointers relating to legal responsibility in case of hostile outcomes. India is at the moment lagging within the institution of a regulatory framework for the testing, licensing and utilization of AI-enabled instruments and units. Countries corresponding to Singapore and the US have notified laws that permit for sure merchandise to be labeled as Software-as-a-Medical-Device (SaMD). However, whereas the draft medical units guidelines issued by the (Union) Government of India in 2016 included SaMDs, the ultimate pointers issued in 2017 don’t. The absence of a transparent regulatory framework hinders deployment in addition to funding in R&D.

Conclusion and Recommendations

Healthcare is a conservative self-discipline, since errors can price lives. For AI to succeed in its full potential, this conservatism should be balanced with pragmatism. To this finish, the general public sector should create a beneficial regulatory ecosystem that addresses the considerations of all related stakeholders in a holistic method. A beneficial regulatory ecosystem is one which considers the incentives and constraints on all stakeholders—from tech firms and their buyers to hospitals, medical doctors and sufferers. Moreover, such a system should be delicate to the character of AI-enabled instruments and the way they differ from conventional medicine and software program. Other necessary points to contemplate embody mental property rights over AI-enabled instruments; medico-legal duty for AI-aided analysis and remedy; regulation across the storage and transmission of affected person knowledge, together with knowledge localisation necessities, safety requirements and transmission protocols; licensing software program as a tool; laws across the high quality and robustness of datasets used to develop fashions; laws for fashions developed utilizing neural networks, the place the relationships between inputs and outputs are successfully a ‘black box’; and import and export laws.

Since coverage is liable for not solely enabling but in addition supporting the nascent ecosystem, it’s needed to guage a broad vary of insurance policies, together with training insurance policies, grant budgets and even mental property frameworks. For occasion, the curriculum in medical faculties should embody acceptable modules on AI to familiarise future generations of clinicians with the benefits in addition to the dangers of utilizing AI in medical settings. For start-ups, taxation coverage should be reviewed and grants be earmarked primarily based on clearly outlined success metrics. The public healthcare system should make investments strategically within the assortment and storage of related knowledge, i.e. monetary expenditure in addition to the coaching of public well being employees. Public our bodies should be inspired to associate with know-how firms and authorities establishments, such because the Medical Council of India and the Indian Council of Medical Research. Finally, the Public Health Foundation of India should spend money on growing a deep understanding of AI, to offer related and knowledgeable coverage steerage over the approaching years.

The personal sector, too, should make a concerted effort to spend money on AI. Recent developments within the area point out that partnerships between know-how, training, healthcare and pharma are the one viable path to the event of AI instruments. Large hospital networks, corresponding to Apollo and Fortis, should discover new methods to collaborate with know-how firms and foyer the federal government for the mandatory regulatory modifications. These initiatives will take years to ship monetary returns and should, subsequently, be performed with a view to long-term, sustainable advantages.


About the Authors

Ashish Airon is Co-Founder of the AI firm Cognitensor.

Ayesha Jhunjhunwala is a healthcare guide.


Endnotes

[1] Data lakes are centralised repositories of enterprise knowledge that include uncooked in addition to reworked knowledge used for duties corresponding to analytics and reporting.

[2]Andre EstevaBrett Kuprel. “Dermatologist-level classification of skin cancer with deep neural networks“. Nature, 2017.

[3]Richards, Sarah. “Early-Warning Algorithm Targeting Sepsis Deployed at Johns Hopkins“. Insight, 2019.

[4] Morley, Dennis. “How Hospitals Are Using AI to Detect and Treat Sepsis“. Healthtech, 2019.

[5]Giacomo Cappon, Martina. “A Neural-Network-Based Approach to Personalize Insulin Bolus Calculation“. Journal of Diabetes Science and Technology, 2018.

[6]Jessica Kent. “Artificial Intelligence Mobile Coach Can Aid Diabetes Care“. Health IT Analysis, 2018.

[7] McKee, Selina. “AI app decreases symptoms of COPD“. Pharmatimes, 2018.

[8] Andre Esteva, Brett Kuprel “Dermatologist-level classification of skin cancer with deep neural networks“. Nature, 2017.

[9] Benjamin J. Ainscough, Erica K. Barnell. “A deep learning approach to automate refinement of somatic variant calling from cancer sequencing data“. Nature genetics, 2018.

[10] Ghalib A. Bello, Timothy J. W. Dawes. “Deep-learning cardiac motion analysis for human survival prediction“. Nature Machine Intelligence, 2019.

[11] Ryan Poplin, Avinash V. Varadarajan. “Prediction of cardiovascular risk factors from retinal fundus photographs via deep learning“. Nature Biomedical Engineering, 2018.

[12]Alexandr A Kalinin, Gerald A Higgins. “Deep learning in pharmacogenomics: from gene regulation to patient stratification“. Future Medicine, 2018.

[13]Adam Lavertu, Greg McInnes “Pharmacogenomics and big genomic data: from lab to clinic and back again“. Human Molecular Genetics, 2018.

[14] Arjun P. Athreya DrewNeavin. “Pharmacogenomics‐Driven Prediction of Antidepressant Treatment Outcomes: A Machine‐Learning Approach With Multi‐trial Replication“. Clinical Pharmacology & Therapeutics, 2018.

[15]Postelnicu, Leontina.”Babylon inks 10-year partnership with Rwandan government“. Mobihealth news, 2020.

[16]Sanofi, Exscientia Ink Up to €250M Deal for Bispecific Drugs Against Metabolic Diseases“. Genetic Engineering and Biotechnology News, 2017.

[17]Dearment, Alaric.”AI drug development partnership between GlaxoSmithKline, Exscientia yields first drug candidate“. MedCity News, 2019.

[18] Stefan Harrer, Pratik Shah. “Artificial Intelligence for Clinical Trial Design“. Rise of Machines in Medicine, 2019.

[19] ROCHESTER, MINN. “Mayo Clinic’s clinical trial matching project sees higher enrollment in breast cancer trials through use of artificial intelligence“. IBM, 2018.

[20] Gregory Yauney, Pratik Shah. “Reinforcement Learning with Action-Derived Rewards for Chemotherapy and Clinical Trial Dosing Regimen Selection“. Proceedings of the 3rd Machine Learning for Healthcare Conference, 2018.

[21] Shu, Catherine. “Amazon’s newest service uses machine learning to extract medical data from patient records“. Techcrunch, 2018.

[22] Lee, Peter. “Bringing together deep bioscience and AI to help patients worldwide: Novartis and Microsoft work to reinvent treatment discovery and development“. Official Microsoft Blog, 2019

[23] Stergiopoulos, S., Fehrle, M., Caubel, P. et al. Adverse Drug Reaction Case Safety Practices in Large Biopharmaceutical Organizations from 2007 to 2017: An Industry Survey. Pharm Med 33, 499–510 (2019).

[24] Schmider J, Kumar K, LaForest C, Swankoski B, Naim K, Caubel PM. Innovation in Pharmacovigilance: Use of Artificial Intelligence in Adverse Event Case Processing. Clin Pharmacol Ther. 2019; 105(4): 954-961.

[25] Chazard, Emmanuel & Ficheur, Grégoire & Bernonville, Stéphanie & Luyckx, Michel & Beuscart, Regis. (2011). Data Mining to Generate Adverse Drug Events Detection Rules. IEEE transactions on information technology in biomedicine : a publication of the IEEE Engineering in Medicine and Biology Society. 15. 823-30. 10.1109/TITB.2011.2165727.

[26]Ministry of Health and Family Welfare, Government of India. “National Health Policy”, 2017.

[27]National Health Authority. “Saal Ek Ayushman Anek, Annual Report of the Pradham Mantri Jan Arogya Yojna 2018-19“.

[28]Boyanagiri, Manaswitha: Boyanagiri, Vamsi Krishna. “Perceptions and experiences of healthcare providers and beneficiaries on the National health insurance scheme of RashtriyaSwasthyaBima Yojana (RSBY) in a Taluk of South Indian State of Karnataka”, Clinical Epidemiology and Global Health, 2018.

[29] Scott Mayer McKinney, Marcin Sieniek, “International evaluation of an AI system for breast cancer creening”, Nature, 2020.

[30] Peng, Lily, “Deep Learning for Detection of Diabetic Eye Disease”, Google AI Blog, 2016.

[31] Landi, Heather, “Health tech funding snapshot—Google joins $50M round in Viz.ai, Augmedix raises $19M and more”, Fierece Healthcare, 2019.

[32]AI and the Health Care Workforce“, American Hospital Association.

[33] John Hawksworth, Richard Berriman, “Will robots really steal our jobs?”, PWC, 2018.

[34] Smith, Yolanda, “Physician Shortage”, News-Medical, 2019.

[35] Alice B. Popejoy, Stephanie M. Fullerton, “Genomics is failing on diversity”, Nature, 2016.

[36] Ayal A. Aizer, Tyler J. Wilhite, “Lack of reduction in racial disparities in cancer‐specific mortality over a 20‐year period”, ACS, 2014.

[37]Billock, Jennifer, “Pain bias: The health inequality rarely discussed”, The Health Gap, 2018.

[38]Pagán, Camille Noe, “When Doctors Downplay Women’s Health Concerns”, TheNewYork Times, 2018.

LEAVE A REPLY

Please enter your comment!
Please enter your name here