Transcription
AI – A brave new world?Life Sciences and Health Care Horizons
Over 500 life science andhealth care practitioners worldwideHow we help Discovery, startup, and growth Research and development Regulatory CommercializationMarkets we serve Pharmaceuticals and biotechnology Medical devices and technology Health care services Hospitals and health care providers Digital health
We work togetherto help you move forward.Digital HealthClinical TrialsAIChinaCell, Tissue, andGene TherapiesAnd more
Life Sciences and Health Care HorizonsDownload our publications
Life Sciences and Health Care HorizonsUpcoming virtual events in 2020Managing Supply Chain DisruptionsWednesday, December 16 at 10:00am ESTThis session will share lessons learned as life sciences and health care organizations have facedsupply chain disruptions as a result of international export restrictions, lockdowns affectingproductions, increased demand, and the restricted capacity caused by the COVID-19 pandemic.Register here.
IntroductionsSpeakersStacy SchultzImogen Ireland General Counsel, Research & Development,UnitedHealth GroupSenior Associate, LondonHogan LovellsJodi ScottFabien RoyPartner, DenverHogan LovellsPartner, BrusselsHogan LovellsPatrice Navarro Dr. Matthias Schweiger Counsel, ParisHogan LovellsPartner, MunichHogan Lovells
What’s all the fussabout?Tackling the elephants in the room
Is AI even a thing?
Is AI even a thing? So, what do we mean when we talk about “AI” today? What’s all the fuss about? “AI” is not new, we’ve used computational modelling fordecades. Why now?
Is AI even a thing?EU definitionU.S. definition[What is AI? US/EU perspective]“Artificial intelligence (AI) refers to systems thatdisplay intelligent behaviour by analysing theirenvironment and taking actions – with somedegree of autonomy – to achieve specific goals.AI-based systems can be purely software-based,acting in the virtual world (e.g. voice assistants,image analysis software, search engines, speechand face recognition systems) or AI can beembedded in hardware devices (e.g. advancedrobots, autonomous cars, drones or Internet ofThings applications).”Broadly defined as the science and engineeringof making intelligent machines, especiallyintelligent computer programs (McCarthy,2007). AI can use different techniques, includingmodels based on statistical analysis of data,expert systems that primarily rely on if-thenstatements, and machine learning.Machine Learning is an AI technique that canbe used to design and train software algorithmsto learn from and act on data. Algorithms maybe is ‘locked’ or ‘adaptive’.
Is AI even a thing?
A brave new world?Case study
Meet the playersPharma[Co]Apps[Co]AI[Co]
Series 1, episode 1‘An unexpected discovery’
Series 1, episode 1PharmaCo would like to review its proprietary compoundlibraries to see whether any of them can be used for a new orfurther medical use.PharmaCo decides that using AI tools could cut the time andcosts of research and development.However, PharmaCo is relatively new to the AI world. So itneeds to seek expertise from AICo, who is working at theforefront of this technology.
Series 1, episode 1AICo has developed a machine learning tool that has combed ahuge amount of publicly available data to build a knowledgemap composed of multiple relationships between targets,diseases, and drugs
Series 1, episode 1PharmaCo and AICo enter into a collaboration agreement.Under this agreement: AICo provides its machine learning tool and expertise; and PharmaCo provides its proprietary data as well as furtherdevelopment facilities for drug candidates.
Series 1, episode 1Good news!They input PharmaCo’s data and AICo’s machine learning tool,identifies five potential compounds for the treatment of a raredisease.AICo’s machine learning tool provides data, which gives theteam a reasonable level of confidence, that these compoundswill be effective and safe.
Series 1, episode 1CLIFFHANGERPharmaCo wants to patent the output, but is it ready yet?PharmaCo wants to know more about the AI’s selection process,but AICo is reluctant to share this information. What willhappen?AICo has looked at the results and is convinced that one of theselected compounds has very little to do with PharmaCo’sproprietary data, and in fact the datasets underlying theinnovation belong to it. How do they decide compensation?
Series 1, episode 2‘What have we created?’
Series 1, episode 2PharmaCo wants to patent the output, but is it ready yet?PharmaCo wants to know more about the AI’s selection process,but AICo is reluctant to share this information. What willhappen?
Series 1, episode 2AICo has looked at the results and is convinced that one of theselected compounds has very little to do with PharmaCo’sproprietary data. The datasets underlying the innovation belongto it. How do they decide compensation?
Series 1, episode 2SHOCK TWISTAICo wants to enter into a new collaboration with anothercompany. What do PharmaCo and AICo do about the fact thatPharmaCo’s data sets have been used to train the AI?
Series 1, episode 3‘So much Data!’
Series 1, episode 3Using AI, PharmaCo realises that one of the drug compounds isactive in a way that is similar to one of its existing drugs,Mediocrezine. While Mediocrezine is pretty effective, it hassome fairly significant adverse events associated with it. Therehave been some anecdotal reports about certain patientcharacteristics that make a patient more or less susceptible toMediocrezine’s more significant adverse events.
Series 1, episode 3PharmaCo has a huge clinical database for Mediocrezine. AsMediocrezine’s development started in the late 90s andcontinued through the early 2000s, the data consents that wereachieved are variable. Additionally, much of the clinical trialwork was conducted in Mainland China.The company would like to run AI on the clinical trial databasefor Mediocrezine to see if it can identify characteristics ofpatients who are more or less susceptible to adverse events.It would then like to take those characteristics, if it identifiesany, and test the new compound against those characteristics tosee if they can end up with a product with similar or betterefficacy but fewer side effects.
Series 1, episode 3CLIFFHANGERCan PharmaCo freely use the data in the clinical trial databasefor another purpose than the original study purpose? If so whatare the limitations which should be taken into account?What challenges do the old informed consent document posesince none of them contemplated this use?Does the concentration of study subject in Mainland China poseany challenges?
Series 1, episode 4‘Let’s run the diagnostics’
Series 1, episode 4Because PharmaCo knows that the development of the newcompound is tenuous at best, it also wants to develop a softwareapplication for physicians to use with Mediocrezine that willhelp them better identify patients that will do well onMediocrezine and ones that will likely do poorly (either due toinadequate effectiveness or unacceptable side effects).PharmaCo has given AICo access to the clinical trial databaseand told them that it wants them to identify the indicators todevelop an algorithm for differentiating between patients.
Series 1, episode 4PharmaCo and AICo work together to develop an algorithm thatcan differentiate between patients though common blood testresults, completion of a questionnaire about family history, Xrays, and an unexpected correlation with educational level.PharmaCo identifies a software development companyAppsRUs to build out the application ‘Ealthy-n-‘Appy and theylaunch it as soon as it’s built. The company does limited testingof the application and does not obtain any regulatory review.They believe it is not rendering a diagnosis and is onlyproviding information to a physician.
Series 1, episode 4SHOCK TWISTUnfortunately, soon after launch of ‘Ealthy-n-‘Appy, theyidentify an increase in adverse event reports and a cluster ofinexplicable deaths.
Series 1, episode 4CLIFFHANGERAre PharmaCo, AICo, and AppsRUs taking the right regulatoryapproach to the application?Does PharmaCo need to do anything in relation to themarketing authorisation of the drug?What are the risks for the legal manufacturer of the software ifcompliance with the regulation is not ensured?
Series 1, episode 4“Have you or a loved one been injured or hurtdue to the application ‘Elthy-n-‘Appy? Youvery likely may have a personal injury lawsuitand be entitled to financial compensation foryour injuries. If you are unsure, you shouldmake the effort to know your rights. ContactAI Injury Law Office today.”
Series 1, episode 4CLIFFHANGERWhich entity is responsible and liable for the software:PharmaCo, AICo, or AppsRUs?Who else is going to have problems with PharmaCo, AICo, orAppsRUs?
Critic reviews of episodes 3 and 4“Wow, AI could save the time and costs of drug discovery. I’mconvinced that AI will replace humans in innovation.”“I felt so bad for PharmaCo and AICo at the end of episode 4. AIshould be held responsible for any errors it makes.”“So many issues, Where to start!”
Series 2?
Are we prepared for AI?
Questions?
Contact usSpeakersStacy SchultzImogen Ireland General Counsel, Research & Development,UnitedHealth Groupstacy.schultz@optum.comSenior Associate, LondonHogan Lovellsimogen.ireland@ hoganlovells.comJodi ScottFabien RoyPartner, DenverHogan Lovellsjodi.scott@ hoganlovells.comPartner, BrusselsHogan Lovellsfabien.roy@hoganlovells.comPatrice Navarro Dr. Matthias Schweiger Counsel, ParisHogan Lovellspatrice.navarro@ hoganlovells.comPartner, MunichHogan Lovellsmatthias.schweiger@ hoganlovells.com
www.hoganlovells.com"Hogan Lovells" or the "firm" is an international legal practice that includes Hogan Lovells International LLP, Hogan LovellsUS LLP and their affiliated businesses.The word "partner" is used to describe a partner or member of Hogan Lovells International LLP, Hogan Lovells US LLP orany of their affiliated entities or any employee or consultant with equivalent standing. Certain individuals, who aredesignated as partners, but who are not members of Hogan Lovells International LLP, do not hold qualifications equivalentto members.For more information about Hogan Lovells, the partners and their qualifications, see www.hoganlovells.com.Where case studies are included, results achieved do not guarantee similar outcomes for other clients. Attorneyadvertising. Images of people may feature current or former lawyers and employees at Hogan Lovells or models notconnected with the firm. Hogan Lovells 2020. All rights reserved.
Download our publications. Life Sciences and Health Care Horizons . A brave new world? Case study . Meet the players . decides that using AI tools could cut the time and costs of research and development. However, PharmaCo is relatively new to the AI world. So it
