This article has been accepted for publication in a future issue of this journal, but has not been

fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JBHI.2020.2970807, IEEE Journal of
Biomedical and Health Informatics
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 7

Short Keynote Paper: Mainstreaming

Personalized Healthcare – Transforming
Healthcare through new era of Artificial
Intelligence
Ketan Paranjape, Michiel Schinkel, Prof. Dr. Prabath Nanayakkara

through financial incentives and increased penalties for

Abstract—Medicine has entered the digital era, driven by data violation of the HIPAA privacy and security rules [4].
from new modalities, especially genomics and imaging, as well as Along with digitized medical records, it is estimated that by
new sources such as wearables and Internet of Things. As we gain
2020, medical knowledge will double itself every 73 days [5].
a deeper understanding of the disease biology and how diseases
affect an individual, we are developing targeted therapies to A doctor would need to spend 29 hours a day absorbing new
personalize treatments. There is a need for technologies like medical knowledge to stay up to date. In other words, we have
Artificial Intelligence (AI) to be able to support predictions for reached the capacity of the human brain and time to follow and
personalized treatments. In order to mainstream AI in healthcare process the new medical knowledge that is being generated and
we will need to address issues such as explainability, liability and
published.
privacy. Developing explainable algorithms and including AI
training in medical education are many of the solutions that can In an era of digital technology, we will be able to increasingly
help alleviate these concerns. tailor medical treatment to the needs of individuals and small
groups of patients. More information will be captured, stored
Index Terms—Artificial Intelligence, Deep Learning, Machine and analyzed to learn how diseases manifest themselves and
Learning, Personalized Healthcare. how patients experience them every day. Combined with a
deeper understanding of molecular science and new methods
I. INTRODUCTION for diagnostics, this development will bring disruptive change
st
21 century healthcare professionals are confronted by many to how we research, develop, approve and pay for medicines, as
technological advancements and large amounts of data. well as how patients and their physicians make decisions about
Physicians and nurses are overwhelmed by data from infusion whether, when and how to treat their illnesses.
pumps, vital sign monitors, laboratory tests, molecular tests,
medical images and all the data that has been recorded in III. NEW SOURCES OF DATA
electronic medical records [1] [2]. Gathering this data and using
As knowledge in medicine, diseases and science grows, high-
it to make an informed and personalized decision poses a
quality data from a wide array of sources can be collected for
unique challenge that has yet to be overcome. New technologies
each patient and can be connected to data from large pools of
such as artificial intelligence (AI) have the intrinsic ability to
other patients for analysis [6] [7]. This enables us to arrive at a
gain insights from large amounts of data from various sources
deeper understanding of disease biology and its expression in
and may be used to solve these problems.
individual patients [8]. Patients are more knowledgeable and
informed, and in the position to demand innovative and
II. DIGITIZATION OF HEALTHCARE
effective treatments. Real-world evidence [9], molecular
An explosion of data and knowledge in medicine, diseases and information generated from next-generation sequencing [10]
science is beginning to impact the healthcare industry, bringing [11], data from wearable devices [12] and mobile apps [13] and
with it a real transformation in care. The Health Information novel clinical trials [14] [15] are increasing our understanding
Technology for Economic and Clinical Health Act (HITECH of health and disease. The regulatory environment needs to and
Act) of 2009 resulted in Electronic Health Record (EHR) is evolving and adjusting for these novel approaches to
adoption to increase from 9.4% in 2008 to 83.8% in 2015 [3] healthcare [16]. The task of unlocking the ecosystem of digital
healthcare cannot be done by anyone alone. As a result, new

There are no sources of funding or sponsorships to be acknowledged in Prof. Dr. Prabath Nanayakkara is with Department of Internal Medicine,
relation this paper. None of the authors declare any conflict of interest. Amsterdam University Medical Center, De Boelelaan 1117, 1081 HV
Ketan Paranjape is a Vice President at Roche Diagnostics, 9115 Hague Rd., Amsterdam, Netherlands (e-mail: p.nanayakkara@amsterdamumc.nl).
Indianapolis, IN 46256, USA (e-mail: ketan.paranjape@roche.com).
Dr. Michiel Schinkel is with Department of Internal Medicine, Amsterdam
University Medical Center, De Boelelaan 1117, 1081 HV Amsterdam,
Netherlands (e-mail: m.schinkel@amsterdamumc.nl).

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JBHI.2020.2970807, IEEE Journal of
Biomedical and Health Informatics
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 7

types of partnerships are forming to ensure we are moving B. Sepsis

towards value-based, personalized patient care [17] [18]. A similar narrative review of AI applications for sepsis was
published in 2019 [37], showing that applications to improve
IV. AI IN HEALTHCARE diagnosis, treatment and prognosis exist already. Many
With the digitization of healthcare, technologies such as AI can algorithms to predict sepsis onset have been developed, with
help us analyze these vast amounts of data to derive insights and encouraging results [38] [39]. However, there are no clear
help with decision making. molecular abnormalities on which new algorithms can be
AI in healthcare is the use of complex algorithms and software trained. The rapid onset and heterogeneous presentation of this
to emulate human cognition in the analysis of complicated syndrome makes it so, that the understanding of
medical data without direct human input. Since a seminal paper pathophysiology remains poor when compared to that of lung
by Sir Alan Turing in 1950 [19], AI has had many advances in cancer. The potential of AI is therefore limited, as unique
Natural Language Processing (NLP) [20], Machine Learning features needed to do adequate predictions are not yet known.
[21], Deep Learning [22], Speech Recognition [23], Virtual Machine learning has the ability to classify in the absence of
Agents [24], and AI-optimized Hardware [25], amongst others. unique features, but to detect conditions like sepsis more data is
Today, AI is already used in healthcare [26] for example to needed because of heterogenous presentation and unique
decrease false-positive results in screening for breast cancer features that are needed in order to provide understanding to
[27] [28], reduce medical transcription costs [29], improve develop new treatments.
physician workflow while relieving and helping to prevent Algorithms can be trained to predict the best possible treatment
burnout [30], robotic surgery resulting in shorter length of on an individual level [40], but can only consider the general
hospitalization and loss of blood [31] and predicting mortality treatments that exist today - antibiotics, source control and
rates of patients with acute heart failure [32]. intravenous fluids. Likely, better treatment options exist, but the
machine learning algorithms are limited by human knowledge
In the past, the most important stakeholder in healthcare, which
at this point in time.
is the patient, suffered from a broad category of diseases which
For AI to be able to provide personalized predictions for
were treated with the same medicines, leaving physicians to
treatment, meaningful data at scale is needed. Clinical trial data,
puzzle over why they worked for some people and not others.
molecular data and general patient data needs to be integrated
Today scientists have begun to understand, target, and diagnose
in advanced predictive models. A broad understanding of
illnesses on an individual level and AI can play a significant
pathophysiology in a certain field is needed in order for AI to
role in this process given its unique capabilities of detecting
become valuable.
subtle disease specific patterns from a wide array of sources,
such as molecular diagnostics, that humans would never
recognize. VI. MAINSTREAMING AI IN HEALTHCARE
As discussed, some disease specific challenges, such as with
V. PERSONALIZED MEDICINE sepsis, hold back the mainstream adoption of AI in certain fields
for now, but there are also some general concerns and
With the use of machine learning applications, a subcategory of
challenges about the adoption of AI in healthcare which have to
AI, that can combine data from all state-of-the-art diagnostic
be addressed at a larger scale.
tests and other resources, there is more potential for
personalized medicine than ever before. A high-level A. Challenges with AI
discussion of two specific fields of medicine will show what AI, Challenges with the introduction of AI in healthcare are
in combination with all these new technologies, can and cannot centered around explainability [41], liability [42] [43] and
do. privacy. Furthermore, the medical educational system for
healthcare professionals will have to undergo a rigorous
A. Lung Cancer transformation.
A 2018 narrative review on AI applications for non-small cell Lack of explainability of AI algorithms is likely to bring about
lung cancer shows that there are already many applications some resistance by the medical community. The more accurate
being tested in this field [33]. Machine learning algorithms can the algorithms, such as neural networks, the less explainable
be used to increase our understanding of important genomic they are. This “black box” phenomenon [43] makes it hard for
pathways in lung cancer, with the use of microarray data [34]. healthcare professionals to get used to working with AI and
Also, machine learning can be used to predict which patient will trusting the algorithm [44]. In the end, physicians still have to
respond to newly developed checkpoint inhibitors [35] or make a final decision and not knowing why you would make a
personalize radiation therapy [36], thereby choosing an optimal certain decision will raise many more issues when a patient is
treatment strategy. A key feature in the success of AI for lung given the wrong diagnosis. Software developers will have to
cancer is that many molecular abnormalities have already been take this into account and prioritize both explainability and
discovered, such as mutations in the epidermal growth factor accuracy. Having explainability will likely also simplify
receptor (EGFR) and anaplastic lymphoma kinase (ALK) [33]. acceptance by the US Food and Drug Administration (FDA) as
These very specific markers provide an excellent starting point mentioned in the recent documents regulating AI [45] and
for algorithms to work from. presumably by medical regulatory agencies around the world.

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JBHI.2020.2970807, IEEE Journal of
Biomedical and Health Informatics
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 7

Then there is an issue with liability. Who is to blame when VII. CONCLUSIONS
something goes wrong? There is no case law about the use of Both personalized healthcare and AI are evolving. As we
medical AI yet. Even worse, the current laws seem to understand more about the biology, diagnostics, and augment
“incentivize physicians to minimize the potential value of AI” medical knowledge with patient data from images, genomics,
[42] as they will only face liability when current protocols are
and medical records, we will be able to identify personalized
not adhered to. New malpractice laws will have to be developed
to specify the liability of all involved parties: healthcare therapies for individuals. As we gain a deeper understanding of
professionals, hospitals, software companies, software how AI works, healthcare professionals will be able to explain
developers and the data collectors. the decision they make with the help of AI tools. With the help
Privacy is another outstanding issue with the use of AI. Vast of technology and regulatory bodies we will be able to resolve
amounts of patient data are needed for some AI algorithms to challenges with liability and privacy. We are well on our way
properly function. Google for example is using 46 billion data to provide personalized treatment strategies driven by AI.
points collected from 216,221 adults’ de-identified data over 11
combined years from two hospitals to predict the outcomes of REFERENCES
hospitalized patients [46] [47]. This raises questions about how [1] J. A. Hazelzet, “Can fuzzy logic make things more clear?,” Crit.
this data is obtained and whether all patients have had a fair Care, vol. 13, no. 1, p. 116, 2009.
chance to decide about the use of their data. [2] “Big Data in Healthcare: Challenges &amp; Promise | NEJM
Catalyst.” [Online]. Available: https://catalyst.nejm.org/big-data-
Lastly, as patients begin to see the benefits of AI and healthcare/. [Accessed: 26-Nov-2019].
proactively use tools like the chatbot [48], physicians will need [3] “Adoption of Electronic Health Record Systems among U.S. Non-
to be aware of limitations of such technologies and care for the Federal Acute Care Hospitals: 2008-2015.” [Online]. Available:
patient accordingly [49]. They will need to be trained in how to https://dashboard.healthit.gov/evaluations/data-briefs/non-federal-
acute-care-hospital-ehr-adoption-2008-2015.php. [Accessed: 25-
effectively use such technologies to their benefit and help ease Nov-2019].
their burden [50]. [4] J. Adler-Milstein, A. J. Holmgren, P. Kralovec, C. Worzala, T.
Searcy, and V. Patel, “Electronic health record adoption in US
B. Resolving Challenges hospitals: the emergence of a digital ‘advanced use’ divide,” J. Am.
In order to alleviate the main concern with explainability we Med. Informatics Assoc., vol. 24, no. 6, pp. 1142–1148, Nov. 2017.
[5] P. Densen, “Challenges and Opportunities Facing Medical
need models that can explain the why, so a physician can Education,” Trans. Am. Clin. Climatol. Assoc., vol. 122, p. 48, 2011.
confidently diagnose a patient with a certain disease. [6] P. Deshpande et al., “Big Data Integration Case Study for Radiology
Explainable AI is a new emerging discipline that is working Data Sources,” in 2018 IEEE Life Sciences Conference (LSC), 2018,
towards making machine decisions transparent, interpretable, pp. 195–198.
[7] S. Tsumoto and S. Hirano, “Healthcare IT: Integration of consumer
traceable, and reproduceable [51] [52] [53]. healthcare data and electronic medical records for chronic disease
The healthcare community needs to be educated regarding these management,” in 2014 IEEE International Conference on Granular
challenges and how to address them and also establish standards Computing (GrC), 2014, pp. 310–315.
and guidelines so a physician and a machine working together [8] M. R. S. Rao and President, “Disease biology in the coming
decade(s).,” Ann. Neurosci., vol. 19, no. 4, pp. 149–50, Oct. 2012.
has the greatest potential to improve clinical decision-making [9] V. B. Katkade, K. N. Sanders, and K. H. Zou, “Real world data: an
and patient health outcomes. opportunity to supplement existing evidence for the use of long-
Medical students, residents, fellows and practicing physicians established medicines in health care decision making.,” J.
need to have knowledge of AI, data sciences, EHR Multidiscip. Healthc., vol. 11, pp. 295–304, 2018.
[10] S. Yohe and B. Thyagarajan, “Review of Clinical Next-Generation
fundamentals and ethics and legal issues concerning AI. Sequencing,” Arch. Pathol. Lab. Med., vol. 141, no. 11, pp. 1544–
Medical schools will need to include them as part of the 1557, Nov. 2017.
curriculum. A staged approach to educating the medical student [11] M. Zhao, D. Liu, and H. Qu, “Systematic review of next-generation
through their journey is recommended [54]. sequencing simulators: computational tools, features and
perspectives,” Brief. Funct. Genomics, vol. 16, no. 3, p. elw012, Apr.
In Jun 2018, the American Medical Association’s House of 2016.
Delegates comprised of proportional representations of every [12] G. Gresham et al., “Wearable activity monitors in oncology trials:
major national medical specialty society and state medical Current use of an emerging technology,” Contemp. Clin. Trials, vol.
associations adopted its first policy on healthcare Augmented 64, pp. 13–21, Jan. 2018.
[13] M. Sharp and D. O’Sullivan, “Mobile Medical Apps and mHealth
Intelligence [55]. Some of the recommendations included Devices: A Framework to Build Medical Apps and mHealth Devices
identifying opportunities to integrate the perspective of in an Ethical Manner to Promote Safer Use - A Literature Review.,”
practicing physicians into the development, design, validation Stud. Health Technol. Inform., vol. 235, pp. 363–367, 2017.
and implementation of healthcare AI; encouraging education [14] E. R. Dorsey, C. Venuto, V. Venkataraman, D. A. Harris, and K.
Kieburtz, “Novel Methods and Technologies for 21st-Century
for patients, physicians, medical students, other healthcare Clinical Trials,” JAMA Neurol., vol. 72, no. 5, p. 582, May 2015.
professionals, and health administrators to promote greater [15] E. L. Korn and B. Freidlin, “Adaptive Clinical Trials: Advantages and
understanding of the promise and limitations of healthcare AI; Disadvantages of Various Adaptive Design Elements,” JNCI J. Natl.
and exploring the legal implications of healthcare AI, such as Cancer Inst., vol. 109, no. 6, Jun. 2017.
[16] “Software as a Medical Device (SaMD) | FDA.” [Online]. Available:
issues of liability or intellectual property, and advocate for https://www.fda.gov/medical-devices/digital-health/software-
appropriate professional and governmental oversight for safe, medical-device-samd. [Accessed: 25-Nov-2019].
effective, and equitable use of and access to healthcare AI. [17] G. S. Ginsburg and K. A. Phillips, “Precision Medicine: From
Science To Value,” Health Aff., vol. 37, no. 5, pp. 694–701, May
2018.
[18] K. A. Teng and D. L. Longworth, “Personalized healthcare in the era
of value-based healthcare,” Per. Med., vol. 10, no. 3, pp. 285–293,

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JBHI.2020.2970807, IEEE Journal of
Biomedical and Health Informatics
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 7

May 2013. using only vital sign data in the emergency department, general ward
[19] “Computing Machinery and Intelligence on JSTOR.” [Online]. and ICU.,” BMJ Open, vol. 8, no. 1, p. e017833, 2018.
Available: https://www-jstor- [40] M. Komorowski, L. A. Celi, O. Badawi, A. C. Gordon, and A. A.
org.iclibezp1.cc.ic.ac.uk/stable/2251299?seq=1#metadata_info_tab_ Faisal, “The Artificial Intelligence Clinician learns optimal treatment
contents. [Accessed: 28-Jul-2019]. strategies for sepsis in intensive care.,” Nat. Med., vol. 24, no. 11, pp.
[20] K. Kreimeyer et al., “Natural language processing systems for 1716–1720, 2018.
capturing and standardizing unstructured clinical information: A [41] A. Holzinger, C. Biemann, C. S. Pattichis, and D. B. Kell, “What do
systematic review,” J. Biomed. Inform., vol. 73, pp. 14–29, Sep. we need to build explainable AI systems for the medical domain?,”
2017. Dec. 2017.
[21] B. J. Erickson, P. Korfiatis, Z. Akkus, and T. L. Kline, “Machine [42] W. N. Price, S. Gerke, and I. G. Cohen, “Potential Liability for
Learning for Medical Imaging,” RadioGraphics, vol. 37, no. 2, pp. Physicians Using Artificial Intelligence,” JAMA, Oct. 2019.
505–515, Mar. 2017. [43] “Artificial Intelligence, Medical Malpractice, and the End of
[22] R. Miotto, F. Wang, S. Wang, X. Jiang, and J. T. Dudley, “Deep Defensive Medicine | Bill of Health.” [Online]. Available:
learning for healthcare: review, opportunities and challenges,” Brief. https://blogs.harvard.edu/billofhealth/2017/01/26/artificial-
Bioinform., vol. 19, no. 6, pp. 1236–1246, Nov. 2018. intelligence-medical-malpractice-and-the-end-of-defensive-
[23] B. Gold, N. Morgan, and D. Ellis, Speech and Audio Signal medicine/. [Accessed: 14-Oct-2018].
Processing. Hoboken, NJ, USA: John Wiley & Sons, Inc., 2011. [44] R. Miotto, L. Li, B. A. Kidd, and J. T. Dudley, “Deep Patient: An
[24] “The Future of Customer Service Is AI-Human Collaboration.” Unsupervised Representation to Predict the Future of Patients from
[Online]. Available: https://sloanreview.mit.edu/article/the-future- the Electronic Health Records,” Sci. Rep., vol. 6, no. 1, p. 26094,
of-customer-service-is-ai-human-collaboration/. [Accessed: 28-Jul- May 2016.
2019]. [45] “Artificial Intelligence and Machine Learning in Software as a
[25] “Machine Learning Benchmark Set with IBM POWER9 and Medical Device | FDA.” [Online]. Available:
GPUs.” [Online]. Available: https://www.fda.gov/medical-devices/software-medical-device-
https://www.ibm.com/blogs/research/2018/03/machine-learning- samd/artificial-intelligence-and-machine-learning-software-medical-
benchmark/. [Accessed: 28-Jul-2019]. device. [Accessed: 09-Dec-2019].
[26] K.-H. Yu, A. L. Beam, and I. S. Kohane, “Artificial intelligence in [46] “Google is using 46 billion data points to predict a hospital patient’s
healthcare,” Nat. Biomed. Eng., vol. 2, no. 10, pp. 719–731, Oct. future — Quartz.” [Online]. Available:
2018. https://qz.com/1189730/google-is-using-46-billion-data-points-to-
[27] J.-J. Mordang, A. Gubern-Mérida, A. Bria, F. Tortorella, G. den predict-the-medical-outcomes-of-hospital-patients/. [Accessed: 01-
Heeten, and N. Karssemeijer, “Improving computer-aided detection Nov-2018].
assistance in breast cancer screening by removal of obviously false- [47] A. Rajkomar et al., “Scalable and accurate deep learning for
positive findings,” Med. Phys., vol. 44, no. 4, pp. 1390–1401, Apr. electronic health records,” Jan. 2018.
2017. [48] M. Bates, “Health Care Chatbots Are Here to Help,” IEEE Pulse, vol.
[28] K. J. Dreyer and J. R. Geis, “When Machines Think: Radiology’s 10, no. 3, pp. 12–14, May 2019.
Next Frontier,” Radiology, vol. 285, no. 3, pp. 713–718, Dec. 2017. [49] A. Palanica, P. Flaschner, A. Thommandram, M. Li, and Y. Fossat,
[29] K. Saxena, R. Diamond, R. F. Conant, T. H. Mitchell, I. G. Gallopyn, “Physicians’ Perceptions of Chatbots in Health Care: Cross-Sectional
and K. E. Yakimow, “Provider Adoption of Speech Recognition and Web-Based Survey,” J. Med. Internet Res., vol. 21, no. 4, p. e12887,
its Impact on Satisfaction, Documentation Quality, Efficiency, and Apr. 2019.
Cost in an Inpatient EHR.,” AMIA Jt. Summits Transl. Sci. [50] “How Chatbots Can Ease Primary Care Burden | Medpage Today.”
proceedings. AMIA Jt. Summits Transl. Sci., vol. 2017, pp. 186–195, [Online]. Available:
2018. https://www.medpagetoday.com/primarycare/generalprimarycare/72
[30] “3 ways medical AI can improve workflow for physicians | American 802. [Accessed: 10-Aug-2019].
Medical Association.” [Online]. Available: https://www.ama- [51] L. H. Gilpin, D. Bau, B. Z. Yuan, A. Bajwa, M. Specter, and L. Kagal,
assn.org/practice-management/digital/3-ways-medical-ai-can- “Explaining Explanations: An Overview of Interpretability of
improve-workflow-physicians. [Accessed: 03-Feb-2019]. Machine Learning,” May 2018.
[31] Medical Advisory Secretariat, “Robotic-assisted minimally invasive [52] M. Wu, M. C. Hughes, S. Parbhoo, M. Zazzi, V. Roth, and F. Doshi-
surgery for gynecologic and urologic oncology: an evidence-based Velez, “Beyond Sparsity: Tree Regularization of Deep Models for
analysis.,” Ont. Health Technol. Assess. Ser., vol. 10, no. 27, pp. 1– Interpretability,” Nov. 2017.
118, 2010. [53] “Interpreting Deep Neural Networks – ShallowMind.” [Online].
[32] J.-M. Kwon et al., “Artificial intelligence algorithm for predicting Available:
mortality of patients with acute heart failure.,” PLoS One, vol. 14, no. http://www.shallowmind.co/jekyll/pixyll/2017/12/30/tree-
7, p. e0219302, 2019. regularization/. [Accessed: 26-Nov-2019].
[33] M. Rabbani, J. Kanevsky, K. Kafi, F. Chandelier, and F. J. Giles, [54] K. Paranjape, M. Schinkel, R. Nannan Panday, J. Car, and P.
“Role of artificial intelligence in the care of patients with nonsmall Nanayakkara, “Introducing Artificial Intelligence Training in
cell lung cancer,” European Journal of Clinical Investigation, vol. Medical Education (Preprint),” JMIR Med. Educ., Sep. 2019.
48, no. 4. Blackwell Publishing Ltd, 01-Apr-2018. [55] “AMA Passes First Policy Recommendations on Augmented
[34] T. Kikuchi et al., “Expression profiles of non-small cell lung cancers Intelligence | American Medical Association.” [Online]. Available:
on cDNA microarrays: Identification of genes for prediction of https://www.ama-assn.org/ama-passes-first-policy-
lymph-node metastasis and sensitivity to anti-cancer drugs,” recommendations-augmented-intelligence. [Accessed: 15-Oct-
Oncogene, vol. 22, no. 14, pp. 2192–2205, Apr. 2003. 2018].
[35] P. Charoentong et al., “Pan-cancer Immunogenomic Analyses Reveal
Genotype-Immunophenotype Relationships and Predictors of
Response to Checkpoint Blockade,” Cell Rep., vol. 18, no. 1, pp.
248–262, Jan. 2017.
[36] G. Amit et al., “Automatic learning-based beam angle selection for
Ketan Paranjape was born in Pune, India, in 1972. He
thoracic IMRT,” Med. Phys., vol. 42, no. 4, Apr. 2015. received the B.S. degree in electronics
[37] M. Schinkel, K. Paranjape, R. S. N. Panday, N. Skyttberg, and P. W. engineering from the University of Pune,
B. Nanayakkara, “Clinical applications of artificial intelligence in India in 1994 and the M.S. degree in
sepsis: A narrative review,” Comput. Biol. Med., p. 103488, Oct.
electrical and computer Engineering from
2019.
[38] R. J. Delahanty, J. Alvarez, L. M. Flynn, R. L. Sherwin, and S. S. the University of Wisconsin, Madison, in
Jones, “Development and Evaluation of a Machine Learning Model 1997. He also received his M.B.A. from the
for the Early Identification of Patients at Risk for Sepsis.,” Ann. University of Oregon, in 2006.
Emerg. Med., 2019.
[39] Q. Mao et al., “Multicentre validation of a sepsis prediction algorithm

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.
This article has been accepted for publication in a future issue of this journal, but has not been fully edited. Content may change prior to final publication. Citation information: DOI 10.1109/JBHI.2020.2970807, IEEE Journal of
Biomedical and Health Informatics
> REPLACE THIS LINE WITH YOUR PAPER IDENTIFICATION NUMBER (DOUBLE-CLICK HERE TO EDIT) < 7

From 1997 to 2016, he was at Intel where he completed roles papers on the use of artificial intelligence with regards to sepsis.
including Vice President and General Manager of Life
Sciences, Technical Assistant and Chief of Staff to Intel’s Chief
Technology Officer, and numerous head of product and
engineering roles for software and microprocessors. From 2016
to 2018, he was a Managing Director at Health2047 and VP at
Akiri Networks, couple of Silicon Valley startups funded by the
American Medical Association. Since 2018, he is the Vice Prof. dr. Prabath W.B.
President of Diagnostics Information Solutions at Roche and Nanayakkara was born in Sri
works on commercializing software to harness the power of Lanka and fled Sri Lanka due to the civil
data, diagnostics and other critical information to support better war in 1989. He received his medical
clinical decisions. degree from the Vrije Universiteit
Mr. Ketan has been a member of the US Health IT Standards Amsterdam in 1995. He has been
Committee Precision Medicine Task Force, AAAS-FBI- practicing as a consultant physician
UNICRI Project on Life Sciences and National and in VU university medical centre, the
Transnational Security, International Telecommunication Netherlands since 2001 and has been
Union’s Global Cybersecurity Working Group and World head of the section acute medicine since 2013. In 2019, he was
Health Organization’s Experts Working Group on eHealth. appointed as full professor in acute internal medicine in
Amsterdam University Medical Center. Prof. Nanayakkara has
led multiple large clinical studies, such as the “PreHospital
Michiel Schinkel was born in Utrecht, ANTibiotics Against SepsIs” (PHANTASi) trial.
The Netherlands, in 1994. He received
his medical degree from the Vrije
Universiteit of Amsterdam, in 2019.
Since, he has been a PhD-candidate in
acute internal medicine, focusing on
improving care for patients with sepsis.
He is author and co-author on multiple

This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.