Ensuring the quality and safety of a medicinal product is both a moral and a regulatory requirement across the biopharmaceutical (biopharma) value chain. Pharmacovigilance (PV) is a crucial mechanism for providing evidence to the biopharma industry and regulators of the long-term safety profile of medicines. Meanwhile, patient support programmes (PSPs) help patients manage their medication and disease outcomes more effectively and, in certain circumstances, can improve equitable access and provide early safety signals. A rising incidence of post-launch adverse event reports (AERs) and increasing expectation of more personalised, preventative, predictive and participatory (4P) medicine has coincided with advances in artificial intelligence (AI) technologies and data analytics. This conjunction increases the potential for post-launch strategies to increase safety, improve equity and enhance patient engagement and experience.
As we highlighted in our 2019 overview report, Intelligent biopharma: Forging the links across the value chain, the pace and scale of scientific innovation are improving patient engagement and experience, creating new business models and transforming the biopharma industry. We concluded that digital transformation, with advanced technologies such as AI, was the next logical step in the evolution of biopharma. This could enable innovation in new products and services and improve customer engagement and process efficiency.1
This is the sixth and final report in our series on the role of AI in driving biopharma’s digital transformation. It explores the potential of AI to improve the last step in the value chain – specifically post-market PV and PSPs (figure 1). The overarching aim of the report is to explore how AI can improve patient outcomes by detecting real-time adverse events (AEs) more efficiently and effectively; and how it can be applied to interoperable health data, generated within PSPs to improve the remote monitoring of patients and deliver safer, more personalised and precise treatment options promptly.
PV is fundamental across the biopharma value chain but is particularly relevant in providing the mechanisms needed to monitor the safety and efficacy of medicines post-launch. PV comprises a set of data-driven and process-orientated activities to detect and identify AEs and report these to regulators to optimise the benefit-risk ratio of health care products.
An AE (adverse event) is defined as “any response to a drug which is noxious and unintended, and which occurs at doses normally prescribed for the treatment of disease or the modifications of physiological function.”2 A serious AE is a life-threatening side effect that causes hospitalisation or prolongation of existing hospitalisation, results in persistent, significant disability or incapacity or causes the death of a patient.3
Biopharma’s PV functions have a legal responsibility to collect, process and report details of AEs and other product safety information to regulators. Over recent decades, increased warnings and awareness about AEs, increasing requirements for better safety documentation and enhanced predrug approval reviews, have made drug safety a top priority for health care professionals (HCPs), patients and regulators.
Most AE reports are self-reported by patients and HCPs to biopharma companies, who then pass them to the relevant regulator. Today AEs can be directly reported to regulators by HCPs and patients. However, handling individual case safety reports (ICSRs) requires significant resources across manual workflow models. Consequently, PV spending is predominantly allocated to case processing.4 A number of global health care trends are affecting the PV landscape and reshaping the PV function (figure 2).
Given the above trends, the regulatory requirement to monitor AEs to ensure patient safety post-launch is becoming especially challenging, driven by the growing complexity of product portfolios, the higher volume and variety of products, and increased public awareness. For example, in 2021, over 2.2 million AERs were submitted to the US Food and Drug Administration’s (FDA) Adverse Event Reporting System, up from 500,000 in 2009 (figure 3).5 Consequently, the global technology-enabled PV market is forecast to grow from $7.8 billion in 2022 to $17.36 billion in 2030, a compound annual growth rate of 10.5 per cent.6
Despite the volume of ICSRs increasing year-on-year, it is estimated that more than 90 per cent of AEs go unreported.7 As the increased availability of data sources, data and information on AEs continue to rise, new challenges are emerging, for example:
The obligation for biopharma companies to report AEs that arise from the unique patient insights and increased data being collected means many biopharma companies are adopting automation and advanced analytics to improve transparency across reporting methods and build trust in their PV systems. This includes creating next-generation digital learning systems that can increase the efficiency and cost-effectiveness of generating richer insights on product quality and patient safety.10
Over the past decade, many biopharma companies have adopted alternative delivery models to increase their PV capacity and efficiency, from full-scale redesign to outsourcing their PV collection and analytical functions (pharmacovigilance outsourcing – PVO). PVO can address talent shortages and knowledge gaps by providing access to customised automation and advanced tools from subject experts. PVO providers can also help biopharma companies navigate the increasing volume and complexity of regulatory requirements. Biopharma’s increasing acceptance and adoption of PVO should stimulate market growth in the coming years and increase the adoption of PV and drug safety software to deliver cost-effective monitoring and reporting workflows, maintain safety data and minimise costs.11
A PSP is a data collection system where the marketing authorisation holder (MAH) exchanges information about the use of its medicinal products with HCPs and patients. Traditional examples include post-authorisation patient support and disease management programmes, surveys of patients and HCPs, patient compliance data or compensation/reimbursement schemes.12
PSPs involve direct interaction with patients to help manage medication and disease outcomes (such as awareness and education, adherence and disease management). A PSP can also, where relevant, provide or arrange financial assistance for patients unable to afford the prescribed medication.13
Patients today have access to more information than ever, which can be empowering but also overwhelming. The volume of data makes it difficult for individuals to identify which sources are relevant and appropriate for their own condition or circumstances. However, PSPs can improve patient outcomes by providing personalised guidance on navigating health care services and an understanding of why treatment is needed. Moreover, emphasising supported self-management can improve disease progression tracking, patient knowledge and treatment adherence.
For many patients managing a chronic disease, the health system appears disjointed and siloed. Pharma companies increasingly see the standardisation and extension of PSPs as a top priority, especially if this can overcome the disjointed components of chronic disease management. PSPs can help to:
Successful PSPs integrate and harmonise treatment timing, in-person care and complicated schedules while providing continued support and education to enable patients to self-manage more effectively. A PSP that embraces the tenets of 4P medicine and educates, empowers, involves and reminds each patient, will improve medication adherence and disease prognosis.
The future success of a biopharma company will be dictated by its ability to improve long-term patient outcomes, including providing wrap-around services for innovative therapies. For example:
While providing patients with PSPs is not new, the evolution of digital technology can increase equity of access and the potential value of PSPs. Moreover, growing evidence demonstrates the value of PSPs.16 Digital approaches offer new opportunities to design and deliver PSPs in novel, impactful and cost-effective ways. The advent of automated health devices can reduce the need for patients to manually input data and help patients understand and manage their health more effectively through continuous monitoring. PSPs can be outsourced to access the operational expertise and technology required to continually scale as the amount of data increases.
The COVID-19 pandemic altered the way patients everywhere access and receive care. The use of direct-to-patient channels such as at-home disease monitoring, medicine home delivery services and telehealth grew dramatically, improving health equity by benefitting patients who are less mobile and/or live in more remote areas.
Importantly, the pandemic raised public awareness of the importance of biopharma innovation and drug safety. The roll-out of the vaccines and extensive media reporting of the benefits and side effects increased awareness of the PV monitoring and reporting processes. These advancements in PV have been facilitated through embracing changes in the role of people, processes and technology while maintaining the focus on quality and compliance. For example, the FDA received over one million COVID-19 vaccine AERs in the first year, and the European Medicines Agency (EMA) received a million AERs from the five vaccines used over the past two years (out of some 868 million doses).17
This increase in AERs has demanded innovative approaches to AE reporting and analysis. The UK’s Medicines and Healthcare products Regulatory Agency (MHRA) enhanced their Yellow Card Scheme using AI to enable the processing of high levels of self-reported AEs, deviating away from the stepwise process via the MAHs.18 During the pandemic, biopharma companies were inundated with large volumes of reports, and in response, the industry quickly simplified PV reporting processes and widened access to patient reporting.
The industry was able to uphold compliance and safety through rapid implementation of new frameworks and technologies – bringing forward planned technology investments by as much as five years. Improved reporting processes, and more proactive public reporting of AEs have helped provide manufacturers and regulators with increased drug safety data. According to IQVIA, the number of AEs reported to the EMA increased by around 50 per cent in 2021 compared to 2020.19
Surveillance, collaboration and standardisation have also been a priority. The WHO manual for COVID-19 vaccine safety surveillance aimed to standardise reporting of both AEs following immunisation (AEFI) and AEs of special interest (AESI).20 While collaboration is improving the safety of patients across the globe, barriers to data sharing such as siloed infrastructure and interoperability challenges and lack of transparency still need to be addressed to gain the full benefits of these partnerships and collaborations.
The building blocks needed to successfully transform the digitalisation of PV and PSPs include supportive leadership and a culture of collaboration and experimentation. However, the main driver is the application of emerging technologies, including AI, blockchain and cloud computing, to improve workflow efficiencies and the capacity of insights derived from this data (figure 4).
When we published our overview report in November 2019, machine learning (ML) and deep learning (DL) were already at the core of most AI technologies used by biopharma. However, we concluded that for biopharma to thrive in a digitally transformed industry, companies would need to:
Over two years later, and with the hindsight of biopharma’s response to COVID-19, companies that made significant digital investments before the pandemic are benefitting from their bold vision as digital transformation has accelerated every part of the life sciences value chain. The rest of this report examines how AI is transforming PV and PSPs and what the future might look like.
Biopharma companies can improve health outcomes by applying AI to health data to identify AEs. Applying advanced analytics to the ever-growing flow of real-world data (RWD) from various sources such as AERs, social media, PSPs and wearable apps can automate end-to-end post-launch PV and drive improvements in patient outcomes. This can also support the move towards increasingly personalised treatments, which will be crucial in improving patient safety, particularly detecting AEs or signalling potential AEs in real-time.
The moment a new therapy enters the marketplace, its impact becomes more complex, requiring effective post-market surveillance. AE reporting is a critical and time-consuming part of ensuring the safe and effective use of medicines within the biopharma industry and is also a regulatory requirement. PV involves the collection, assessment and reporting of AEs, or single case processing and the continuous communication, interpretation and monitoring of product benefit-risk profiles to enable signal detection and benefit-risk management.
AI can help address both areas, including providing new opportunities for PV. For example, ML algorithms can be trained to extract and classify information from incoming AERs, using structured, semi-structured and unstructured data (figure 5). AI-extracted and classified information is then submitted to a drug safety specialist for review and confirmation or correction. This ‘augmented intelligence’ of AER processing serves as input for subsequent ML rounds, improving the algorithms over time, increasing consistency and reducing errors. These advancements have the potential to deliver a significant efficiency boost to the current PV operating model.22
AI provides a scalable and adaptable solution for handling the growing case volume and diverse types of incoming data formats effectively. NLP supports the end-to-end transformation of PV by enabling the analysis of multi-formatted AERs, including mass unstructured data littered with colloquialisms and non-technical terminology. The improved effectiveness and consistency of automated AE reporting allow better regulatory compliance by ensuring the timeliness and accuracy of submissions (case study 1).23
PV workflows deliver vast volumes of reports from many sources, including call centre feeds, emails and regulatory AE reports. These generally require coding into a standardised format for processing using the Medical Dictionary for Regulatory Activities (MedDRA). Most of this coding is manual and time-consuming as only when the verbatim is an exact match with a MedDRA term is coding automatic.
A Linguamatics NLP-based AI platform delivers high-value knowledge discovery and decision support from free text reducing the burden of manual coding by traditional methods. The NLP platform offers flexibility, scalability and data transformation power to analyse unstructured data. The NLP solutions read the required documents to understand the many ways a single concept can be represented linguistically and then extracts the required information.
Linguamatics worked with CSL Behring to develop an NLP workflow that doubled the level of auto-coding of AEs from 30 per cent to over 60 per cent. The use of NLP reduced the manual time needed by 50 per cent, improved coding consistency and reduced risk for case processing and medical evaluation.24
The increasing amount of data social media is generating can also be used to detect safety signals. However, colloquial terminology, including emoticons and duplicate reporting, such as parallel posting on multiple platforms, makes it challenging to mine social media data and extract meaningful health-related information.25 AI tools can address this challenge (case study 2).
In PV, monitoring social media as a source for potential AEs is challenging and costly due to the high volume of data and the need to extract meaningful, timely insights from text containing non-traditional written communication like acronyms, emojis, hashtags, images and slang. To counter these challenges, Novartis created an NLP model using Amazon SageMaker to analyse social media posts automatically and, based on the content and context, predict whether they contain reports of AEs.
AE Brain monitors social media channels for mentions of potential AEs, assesses them using sentiment analysis and flags messages of interest to human researchers for review. Sixty per cent of AEs are processed by AE Brain directly as a triage mechanism, enabling humans to focus on a smaller number of significant events of interest. Currently, AE Brain processes around 15,000 messages per week, capturing far more data than a human team could review, and increasing the overall quality of Novartis’ drug monitoring.26
Another value-adding use case is applying NLP to datasets such as free text in social media, news articles and literature to detect unexpected benefits of a pharma product. This approach provides an opportunity for PV to improve patient care while contributing to the top-line revenues of a company.27
As the number of reports and varied data sources increases exponentially, the infrastructure to cope with the growth in large and diverse datasets will need to improve. Likewise, as health care providers continue to evolve from reactive to more proactive-preventive care, they are looking to digital transformation to provide actionable data.
The rising cost of PV tasks has led many companies to outsource these to specialist companies. This is especially important given the steady rise in PV cases. Regulators also require HCPs to report more incidents and encourage patients to share their AE stories via chat groups and social media. The intention is to provide industry bodies with more in-depth trends across treatment categories and patient populations.
Intelligent automation and predictive analytics can be deployed to capture and translate AE data from multiple channels. Using these integrated data sets to conduct deep analysis and identify critical signals will enable biopharma to transform their entire PV workflow to become more efficient and effective. The current generation of PV platforms augments the best data analytics tools with human science expertise. They also offer a variety of automation features, including front-end automation and NLP, making it possible to integrate data into safety databases with minimal human intervention required.
NLP and automation tools can identify relevant information to build clinically robust auto-narratives, eliminating-labour intensive data entry. AI and ML can increasingly identify patterns within structured and unstructured narratives, negating the need for routine reviews of single cases and manual identification and validation of signals.28 Benefits include reducing errors related to manual data entry and anonymised data, reducing time and labour costs and increasing speed in delivering information. Consequently, several companies have developed innovative platforms, for example:
Voice data is becoming ubiquitous and expanding the avenues for detecting previously unidentified AEs and safety insights. PV teams who utilise voice data effectively arm safety teams with the tools to secure patient well-being and increase compliance and product prospects. Biopharma companies can harness safety technology to remove as much as 60 per cent of the ‘noise’ from their safety risk identification workflow. Virtual agents can also enable coverage in multiple languages and monitor AE reports around-the-clock (case study 3).
IQVIA seized the opportunity presented by the pandemic to implement AI technologies to support automated AE report intake and retrieval and used Robotic Processing Automation (RPA) bots to support automated case processing. IQVIA’s Vigilance Detect (powered by AE Tracker®) uses cloud-based technology combined with advanced analytics, including AI and ML. It processes global regulatory intelligence, legal/ethical guidance, product quality information and safety data on an integrated platform to increase visibility across the compliance landscape.31
Situation – A biopharma client needed to monitor AEs generated through pharmaceutical sales representative call notes in its customer relationship management system to address the challenges faced by global drug safety professionals in detecting AEs in user-entered notes.
Solution – Detect was used in a global system of an oncology brand to detect AEs for 12 months of sales representative call data. PV analysts validated the triggered AEs to identify client drug safety issues. The client employed the FDA-recommended random sample audit methodology to ensure the system was error-free.
Outcome – Of the 23,000 call notes, 109 records with AEs were identified, and the systems and processes deployed passed the quality audit with 100 per cent accuracy, successfully detecting all potential AEs.32
While biopharma companies have been providing PSPs for many years, the evolution of analytics, data and technology has increased the potential for these programmes to increase equitable access and improve patient engagement, experience and outcomes. AI can be used as part of a PSP to track and engage with patients remotely and intervene with personalised treatment options in a timely manner. Reliable, interactive and personalised PSPs can improve adherence to medication prescriptions and help manage complex and long-term conditions more effectively.
Pharma companies increasingly need to demonstrate value to payers beyond clinical efficacy. This is particularly true for next-generation cancer therapies involving a complex care regime emphasising quality of life and patient experience. Chronic diseases often have a long-term burden of care, and patients require information to self-manage. A well-designed PSP, enabled by digital technologies, can address these challenges and should include the criteria outlined in figure 6.
AI can help across many of the elements outlined, especially if underpinned by a cloud-based platform based on open data approaches that standardise core data to improve interoperability and allow seamless integration of sensor data and user applications. Moreover, as the industry pivots towards outcome-based reimbursement models, the importance of PSPs will continue to grow requiring flexible solutions that are scalable and adaptable with fast roll out capabilities (case study 4). We have identified seven post-launch activities where AI-enabled PSPs can significantly influence patient outcomes (figure 6).
ConvergeHEALTH aims to help provide solutions to many challenges faced across the health care ecosystem by delivering personalised experiences at scale. It is a flexible platform that offers consistent engagement with patients across multiple channels and the care continuum, helping strengthen HCP-patient partnerships. It also enables adherence by educating, engaging and supporting patients and supports care coordination for each patient’s care network.
Situation - A top global pharmaceutical company’s digital health organisation wanted to provide a holistic solution based on patient-centric engagement to deliver an online platform for patients and HCPs to self-manage the services of access programmes.
Solution – The PSP core is used as a base for a multi-country roll-out (six countries), allowing for a flexible, scalable and extendable solution that delivers a rapid, agile delivery – from blueprint workshops to all functionality within seven months.
Outcome – The solution roll-out led to many positive changes and results, including:
Remote patient monitoring (RPM) using wearables is crucial in empowering patients to self-manage their condition. The COVID-19 pandemic necessitated a shift in care delivery, rapidly accelerating the adoption of RPM solutions. While in-person visits provide a valuable touch point for patients, they are often episodic. In contrast, RPM continuously collects data and information on the progress of an individual’s condition. This is particularly important when treating complex conditions sensitive to changes in medication.
Wearables are increasingly more patient orientated with improved battery life and capacity to remotely monitor health measures for extended periods, such as vital signs (oxygen saturation, blood pressure and blood glucose) and physical characteristics (activity levels, walking speed, step length and asymmetry).34 Pharma’s adoption of AI-enabled digital technologies alongside PSPs, means the wealth of objective and longitudinal data collected can be analysed quickly and processed to provide personalised health insights, facilitate increased patient engagement, anticipate deteriorating health and decrease the costs of care for patients and HCPs.35
The application of AI to PSP-generated health data can be used to:
As the digital infrastructure and competencies mature, and RPM is used more widely in clinical trials, biopharma companies are uniquely positioned to increase RPM in the post-marketing space. The more data available to the AI algorithm, the more it can learn and improve its pattern recognition capabilities, generating more accurate predictions and improved analysis.
Medication adherence is critical in effective disease management, especially for chronic conditions. However, engaging patients to self-track their medication is challenging and often wanes over time, with approximately half of all patients not taking medication as prescribed.36 A British Medical Journal review into the nature and prevalence of non-adherence reported that four per cent of hospital admissions were caused by medication non-adherence. Almost all of those were considered preventable.37 Higher adherence rates may also aid biopharma by reducing waste and increasing the percentage of medication used.
There are many causes of non-adherence, but they fall into two overlapping categories, intentional and unintentional.
Intentional non-adherence occurs when the patient decides not to follow the treatment recommendations.
Unintentional non-adherence occurs when the patient wants to comply but is prevented from doing so by barriers beyond their control, including financial and physical constraints, and difficulties in understanding instructions.
Tailored solutions are needed to overcome the complex and varied barriers to medication adherence. The role of the HCP and PSPs is critical in providing equitable access to information to help patients make decisions in accordance with legal and professional codes of conduct.38
Pharma can deploy AI across a population of PSP participants to predict patients at higher risk of non-adherence. Factors such as age, costs, employment status, ethnicity, gender, medical history and prognosis, postcode and the complexity of the dosage regimen all influence the likelihood of a patient’s adherence to their prescribed regimen.39 If a patient is identified as having a high risk of non-adherence, they can be targeted for proactive, tailored prophylactic interventions (case study 5).
New York based AllazoHealth is a health care technology company that uses the power of AI to optimise patient outcomes. AllazoHealth’s AI platform can predict a patient’s adherence risk at a given time for multiple behaviours and accurately forecast how much a specific intervention would change the risk of non-adherence.
One pharma client faced the challenge that an estimated 50 per cent of patients stopped taking their medication within the first 30 days of treatment. AllazoHealth helped patients within the first 30 days of treatment through a customised, multi-channel programme designed to increase patients’ days on therapy and help them stay on track with their medication. These patients adhered to the treatment for 4.6 times longer than patients that had not been supported by the AI platform. Additionally, emails powered by AI were 4.5 times more effective than randomly assigned targeting, resulting in increased engagement in the PSP.40
Digital technologies provide an opportunity to build more equitable PSPs that reach new patient demographics and provide clinically relevant interventions. To optimise this opportunity, the design process needs to build on data and patient profiles from the target population and apply behavioural design concepts, processes and techniques, including:
Some of the digital technologies that biopharma can use to support patients remotely include:
Accelerated by the pandemic, chatbots are increasingly used as part of a PSP to answer questions relating to medication and side effects or simply for reassurance. Patients can create a personalised profile to enable chatbots to deliver a bespoke experience to help patients stay on track. Research shows that simplistic AI-powered technology interventions can be crucial in low medication adherence among more vulnerable and marginalised patient populations.47 From answering questions relating to medication to addressing reasons for non-adherence, AI-enabled chatbots can provide personalised information to benefit the patient.48 Examples include:
Patient outcomes can be impacted by a lack of trust or understanding of prescribed medications. PSPs should therefore be designed to cover a multitude of concerns, including emotional issues such as fear surrounding medications and the concern of side effects. Multicomponent interventions that include attitudinal, educational and technical aspects to modify and enhance patient medication-taking behaviour, integrated across PSPs, display the most promising results in maintaining long-term medication adherence.52
The Deloitte report in 2021, Overcoming biopharma’s trust deficit, found that trust in biopharma companies is critical if companies are to gain and retain patients and improve health outcomes. Consumer trust also gives biopharma the incentive to innovate and invest in new life-saving therapies.53 Deloitte found that participants in the research trusted HCPs (doctors’ offices and physician groups) the most. In contrast, biopharma companies ranked near the bottom of the list of eight options.54 Deloitte identified four trust-building signals:
AI-enabled PSPs provide a strong opportunity to improve patient trust in biopharma across these four signals and give patients an amplified desire to maintain compliance with their treatment if the benefits are properly articulated and understood by patients, caregivers, payers and HCPs. Moreover, the digital technologies that allow biopharma companies to design and build a tailored patient engagement experience are the same technologies that can drive better HCP engagement and education. Pharma companies need to involve HCPs in developing programmes and supporting assets to drive awareness and ensure that the solution meets the needs and expectations of all stakeholders.
The need for better patient engagement and experience is generating new business models. Over the next few years, patient-centric, co-created support services will make patients equal partners in decision-making and help biopharma companies deliver safer, more personalised health outcomes. An AI-supported PV system that enables patients to identify and report AEs directly and the rise of AI-powered patient-centred PSPs will transform biopharma’s relationship with patients and deliver improved patient outcomes.
There are several challenges to realising the patient-centric future enabled by adopting AI-enabled PV and PSPs. Several steps must be taken to turn these challenges into opportunities (figure 7).
While AI models rely on data to train and learn, improvements in the infrastructure that algorithms run on, including applications, hardware and software, are key factors enabling the utility of AI.56 In the future, data collected, processed and used in real-time by innovative medical devices will be biopharma’s new ‘currency.’ As such, a key differentiator for companies will be how they can spend and earn this currency by generating insights and evidence from multiple, reliable data sources.
However, this will depend on investment in a robust digital infrastructure as AI cannot be imprinted onto an existing siloed digital system. Biopharma companies should prioritise developing the right technology assets to establish a cohesive and interconnected infrastructure with a high level of connectivity and interoperability that supports secure and transparent data exchange. Digital transformation will also impact business models, the development of new products and services, and transform how companies engage with HCPs, patients and other customers.
AI-powered post-market PV and PSPs have the potential to increase the connectivity, effectiveness, efficiency and resilience of the system.57 The combination of predictive data analytics, improved infrastructure (such as 5G connectivity) and novel diagnostics increasingly become more embedded and support a more preventative model with a focus on value (quality of care and outcomes), not the volume of care delivered. These changes will enable HCPs to offer more patient-centred care and spend more time communicating and providing compassionate care.58
The deployment of AI will also require significant changes to roles and responsibilities across the post-market PV value chain. Biopharma employees will need to develop new skills and talent and take into account the increasing volume of remote work. Many biopharma companies are consequently re-evaluating their workforce model. This trend could cause talent redistribution and create a need for up-skilling and capacity-building while helping employees stay responsive to changes. The next generation of biopharma PV and PSP talent will need to be agile, digitally literate and open to continuous learning as part of their career development.59
Moreover, to realise the benefits of AI, the skills and talent required by biopharma companies will need to include advanced analytical cognitive and digital skills, data scientists and software engineers who understand how to design a digital product and solution that meets patient needs. Therefore, there is a need for skilled interdisciplinary leaders to share learnings and support new business and operating models. The talent shortage needed to support digital transformation initiatives is one of the most significant barriers, so expanding and upskilling talent should be a top priority for the C-suite.
PV and truly personalised PSPs are dependent on trust. As patients become increasingly comfortable with virtual care and RPM through wearables and apps, their service expectations will increase. The ability to connect with their HCP on their terms and continuously share the information they choose will be increasingly important in empowering patients to become more active participants in their own care and support the shift to preventative approaches.
Biopharma’s increasing access to personal data generated through PSPs comes with a greater responsibility to strike a balance between protecting individual privacy and utilising the data to improve patients’ health and quality of life.60 Companies who demonstrate active steps to protect patient data will gain a competitive advantage while empowering individuals to understand their own data. They are also likely to continue to build greater trust with patients. While a PSP alone cannot drive differentiation and increase trust and engagement, programmes that are well-designed, interactive and tailored to personal preferences, can.61
As biopharma embraces AI-powered digital transformation across the post-market ecosystem, the amount of patient data in the hands of these companies will exponentially increase. This data needs to be responsibly and securely handled across an integrated data network. This requires biopharma to include ethical considerations into the design, build and deployment of AI-powered systems. This also includes testing and remediating systems that unintentionally introduce bias and treat users unfairly.
Over the next few years, patient-centric, co-created experiences will evolve to make patients more equal partners in decision-making throughout their care pathway, helping biopharma deliver better, more personalised outcomes. However, genuine patient centricity means understanding the patient’s lived experience of their condition – what the individual patient values and needs, and what is most likely to result in a positive health care outcome.62 Implementing AI in PSP design provides an opportunity to view patients’ health holistically. As biopharma companies capture each patient’s unique clinical history, socioeconomic factors and previous experiences, they can create increasingly personalised solutions, providing a seamless delivery of ‘what customers want, where they want it, and when they want it.’
AI provides the opportunity to predict, at a personal level, a patient’s disease trajectory and recommend treatment or highlight the need for potential intervention while improving patient engagement. Research shows there is an expectation from society for this kind of approach, with 83 per cent of patients saying it is important that providers know them personally, beyond their health record.63 This increased level of personalisation has been shown to improve patient engagement and outcomes and increase the level of trust between providers and patients.64 Examples include Orion Health’s AI Engage platform which provides personalised, relevant educational information for patients, encouraging them to actively contribute to their health management by sharing their health information and improving ease of access to their HCPs.65
Many aspects of drug delivery can be personalised in biopharma, from the tailoring of dose to its administration, including frequency. This can lead to better therapeutic outcomes and decreased AEs.66 When combined with AI, a personalised drug delivery regimen can create an interactive open feedback loop between the patients’ needs and the treatments. Twin Health has created a Whole-Body Digital Twin, a dynamic, digital, AI-enhanced representation of metabolic function derived from personal health data points. It monitors daily activities and personal adaptation.67 Their Twin Service provides each user with personalised and precise activity, medication and sleep management guidance.68
The development of digital devices to support patient participation in a PSP should include input from end-users to assure ease of operability and their engagement in the shift from data entry to data quality assessment. Selecting outcomes that provide meaningful business values and achieving them in a specific timeframe will drive PSP professionals, HCPs and patients to adopt these AI capabilities. AI-enabled platforms allow HCPs to interact physically and digitally concurrently. Importantly empowering patients includes explaining how AI is used in jargon-free language and providing use-cases.
Engaging with patients from the beginning of the PSP development process can significantly improve outcomes. The National Institute for Health Research (NIHR) aims to bring patients and life science companies together earlier in the research and development processes to instil a culture of partnership between the biopharma industry organisations and patients, which can be expanded to the development of PSPs.69 With greater transparency, biopharma has scope to go beyond co-design to co-creation, where patients are involved throughout development, testing and roll-out. Creating solutions that patients are invested in helps to build advocates for the PSP and ultimately scales a biopharma company’s ability to make an impact through patient partnerships.70
By collecting information on AEs and acting in response, regulators aim to protect the public from emerging safety issues throughout a treatment's life cycle. In January 2021, the FDA released the AI-based SAMD Action Plan.71 The FDA acknowledges that one of the most significant benefits of AI/ML is its ability to learn from real-world use and experience to improve its performance. The FDA has declared its commitment to support a patient-centred approach and emphasised the need to be transparent about the functioning of AI-based devices to ensure users understand the device’s benefits, limitations and risks.
Similarly, in April 2021, the European Commission published its proposal on AI Regulation, introducing a comprehensive, harmonised regulatory framework for AI with significant turnover-based financial sanctions. For life sciences, the AI Regulation is designed to complement and work alongside several existing legal frameworks, particularly the product safety / CE regime and data protection under GDPR. More specifically, the provider is expected to design the system to ensure:
The application of RWD to create adaptive labels brings new opportunities and challenges to the regulatory relationship. Intelligent automation of pharmaceutical labelling can increase the speed and accuracy of labelling compliance as the complexity and expectations of labelling regulations continue to evolve. Between January 2017 and September 2019, 14.9 per cent of drug recalls by the FDA were due to labelling issues.73
In the future, AI will compare any number of countries’ regulations simultaneously and adjust before non-compliance occurs, helping to ensure resilience against the ever-changing regulatory compliance landscape. This faster updating of new requirements and other label adaptations will enable critical information to be processed and relayed to patients quickly, increasing transparency between the industry and patients.
Transparent and collaborative communications between regulators and biopharma companies about the deployment of AI-enabled technologies in post-market surveillance and PSPs will be essential in maintaining regulatory compliances. Companies will need to report serious incidents and malfunctioning, taking appropriate measures or even withdrawing the AI system when it presents a risk to health, safety, human rights or a public interest. In addition, transparency is essential to ensure that the individual is aware of being exposed to an AI application. Nevertheless, new data-driven approaches will enable biopharma to work more collaboratively with regulators to balance risk and create new evidence frameworks for PV with blockchain-like technology used to verify the origin of data submissions.
Post-launch, biopharma companies will need interoperable health data to track and engage with patients remotely and intervene with personalised treatment options at the right time. This requires integrating and analysing patient data in real-time to determine when an intervention is needed. Applying AI to monitor the safety of patients will be critical, particularly in detecting potential AEs proactively and in real-time. Insights into RPM beyond safety will also be important, and compliance monitoring is an area many biopharma companies and start-ups are targeting. AI-enabled PSPs will transform biopharma’s relationship with customers improving enrolment, adherence and retention, delivering improved patient outcomes.
The Deloitte Centre for Health Solutions is the research arm of Deloitte’s Life Sciences and Health Care practices. We combine creative thinking, robust research and our industry experience to develop evidence-based perspectives on some of the biggest and most challenging issues to help our clients to transform themselves and, importantly, benefit the patient. At a pivotal and challenging time for the industry, we use our research to encourage collaboration across all stakeholders, from pharmaceuticals and medical innovation, health care management and reform, to the patient and health care consumer.