How a Process Orientation Contributes to Success with AI
Image: Depositphotos
It is well known that artificial intelligence (AI) enables better, faster and more automated decisions. Indeed, it has been proposed that AI is driving a resurgence of interest in redesign business processes.[i] That’s partly due to the ability of certain AI tools, such as robotic process automation (RPA) which when combined with machine learning as “intelligent process automation,” can automate information-intensive processes. It has also been argued that AI fits well into improvement methods such as Lean Six Sigma and can be applied at each stage of the so called DMAIC process (Define-Measure-Analyze-Improve- Control).[ii] Note that Six Sigma and Lean Six Sigma are highly codified and structured methods of process improvements which have a strong bias towards incremental improvement within organizational boundaries. The integration of AI into process improvement may have the potential to reignite interest in more major change – targeted at large enterprise processes – perhaps even reengineering.
While there has been a fair amount of attention on how AI is helping companies to improve process performance, there has been less consideration of how a process orientation enables greater success with AI. Adopting a process orientation can help address organizational, cultural and process challenges – which are the biggest roadblocks to success in deploying AI.[iii]
Process orientation principles involve much more than just modeling. It includes attention to customer experience, measurement and governance. So, in addition to modeling the end-to-end process, it encompasses a focus on the customer, a shift in measurement from concentrating mainly on financial factors to considering the timeliness and quality of services provided and governance of the end-to-end process – with particular attention to collaboration.
This article considers how a process orientation may benefit redesigning the oncology clinical trial process which has not changed much in several decades. Why consider the oncology clinical trials process? It is complex, highly regulated and has challenges in obtaining standardized, high-quality data. Next, unlike many other new product development processes, millions of lives are at stake. Then, it requires collaboration across departments inside the organization as well as with external parties. Finally, AI can make a needed difference as the oncology clinical trial process has underperformed all other therapeutic areas as the table below indicates.[iv]
From a business process perspective, there are 5 phases in an oncology clinical trial: design, recruit, conduct, report, and file for approval. As it can cost over a billion dollars and take over a decade of effort to bring one new cancer medication to market, the potential for AI to improve the oncology clinical trial process is immense. A process orientation enables big pharma to compress the time from development to approval and make the process less burdensome on patients by eliminating nonvalue added steps and updating outdated policies.
An overreliance on pilots and proof of concept projects is one of the pitfalls to overcome in deploying AI. Many AI initiatives are too small and too tentative. They rarely get to the stage of adding economic value—being deployed on a large scale.[v] A process orientation can make a significant difference in this regard. Modeling the end-to-end process at the enterprise level and discussing the issues and disconnects can set the stage for a systemic view of the end-to-end oncology clinical trial process, rather than just one-off projects. It is then possible to go beyond simple, incremental improvement and envision a fundamental redesign of the entire process through technology. But it’s not just about technology. Outdated policies, business rules and metrics are also viewed in a new way via the process lens.
Now consider using AI to improve clinical trial recruiting. Process orientation can have a positive impact on recruiting in oncology clinical trials. In oncology, over 32 % of all terminated trials are due to low accrual rates – so recruitment is a major challenge. The opportunity here is to match clinical trials with eligible cancer patients.[vi] AI can analyze huge datasets of health records rapidly and use algorithms to identify potential participants who meet specific trial criteria with accuracy.[vii] For example, Deep6AI has
developed software that helps pharma companies to scan electronic medical records (EMR) data to understand the feasibility of running a specific clinical trial, identify eligible patients for clinical trials, and generate real-world evidence. [viii]
As awareness of clinical trials continues to be challenging, AI opens new channels for intelligent, targeted trial awareness and outreach.
A process orientation would emphasize looking at the process from the outside in (the patient’s perspective) as well as the inside out (the company’s point of view). This involves examining the entire patient journey in an oncology clinical trial, which would reveal several disconnects. For example, the informed consent form (ICF) used in the enrolment sub-process is typically written in technical language and difficult for patients to understand. AI could then be deployed to draft a patient friendly ICF. Then, the screening process is frequently burdensome on patients with lengthy visits to the trial site. AI could be deployed to compress the time needed for trial visits during screening.
Clinical trial sites are often burdened with administrative tasks in participant screening. Generative AI chatbots or virtual assistants could be deployed to ask questions based on the trial’s criteria, collect basic data from potential participants, and even schedule preliminary appointments. This approach can not only reduce the workload of the research team, but also improve patient experience.[ix]
A process orientation can also have a positive impact on the conduct of oncology clinical trials. Wearables and AI techniques could be used to collect data, thereby facilitating data collection. Deep learning models can be used to analyze data and synchronously log relevant events, generating relevant diaries and AI can be deployed to autofill forms needed for reporting.[x] AI can also reduce the number of patients needed for a trial by creating digital twins of patients in clinical trials. Then researchers can use the twin to predict how the same patient would have progressed in the control group, compare outcomes and thereby potentially reduce the number of control patients needed by over 20%. [xi]
Process orientation can have a positive impact on measuring performance in clinical trials. The current focus is on patients’ overall survival and progression free survival. Other key performance indicators (KPIs) such as number of adverse events, drop out rates, and variance to budget are typically measured. While these metrics are indeed important a process orientation would lead organizations to measure other factors that are important to patients – such as timeliness and quality of care. Intelligent KPIs, powered by AI have the potential to measure factors such as:
Visit frequency and duration
Patient satisfaction
Tests done right first time
Issue identification and resolution time
On-time performance by site
Process orientation would also highlight opportunities and issues in cross functional collaboration. Oncology clinical trials demand close collaboration among the principal investigator, clinical nurses, the Pharmacokinetics (PK) nursing team, various testing teams such as the CT (computed tomography) scan team and pharmacy. Process modeling and measurement would identify scheduling issues and delays in hand-offs. Then, AI can create more connected, coordinated systems and thereby help break down silos by enhancing communication, customization, and coordination – both inside and outside the organization.[xii]
It is difficult – perhaps even impossible to make an enterprise-level impact with one project at a time. A process orientation encourages a view across workforces, processes and technologies and requires close attention to:
· Forming cross functional teams and oversight steering teams
· Defining data ownership and assuring transparency
· Encouraging collaboration across groups
· Paying close attention to change management
· Having common objectives and KPIs
While deploying AI is challenging and requires large volumes of high-quality data in standard format and people with the right skills, the biggest roadblocks to success are arguably organizational and cultural challenges. Adopting a process orientation can help in this respect, especially in highly regulated and risk averse sectors such as pharmaceuticals. But it does require shifting management attention. Instead of just focusing on departments, companies with a process orientation will also focus on what customers expect and the value creating flow of work that crosses organizational boundaries.
Consider the following self assessment questions to determine the scope of effort needed to adopt a patient centric process orientation.
1. Does your company begin the strategy process with an understanding of customer (i.e. patient) expectations?
2. Is your strategy expressed in terms that people at various levels throughout the organization can understand?
3. Do your leaders understand how workflows across the traditional organizational boundaries? Do they “know the business” in this context and at the right level of detail?
4. Are there visible rewards and recognition for those people who are instrumental in improving the performance of the firm’s large, cross-functional processes?
5. Are your leaders more concerned about reporting relationships and authority than on the flow of work to create value for customers (i.e. patients)?
6. Is it common that executives and managers view the business with a functional bias?
7. To what extent is there a shared understanding of the business process view at the top team level?
8. Do people generally view processes on a micro level as procedures?
9. Do leaders focus mainly on actual-to-budget metrics to assess progress?
10. Are quality and timeliness metrics lacking?
Don’t underestimate the needed effort. Shifting management attention is hard work. However, AI is fast becoming a pervasive technology and organizations that learn how to use a process orientation in deploying AI are likely to enjoy greater success.
[i] Davenport, Thomas H., Matthias Holweg, and Dan Jeavons. "How AI Is Helping Companies Redesign Processes." Harvard Business Review, March 2, 2023.
[ii] Holweg, Matthias, Thomas H. Davenport, and Ken Snyder. "How AI Fits into Lean Six Sigma." MIT Sloan Management Review, November 9, 2023.
[iii] Davenport, Thomas H., and Nitin Mittal. "How CEOs Can Lead a Data-Driven Culture." Harvard Business Review, March 23, 2020.
[iv] There has been some improvement in oncology clinical trials as reported by "Shifts in the Clinical Trial Landscape." Nature Reviews Drug Discovery Volume 23 | April 2024 | 238–239 |
[v] Davenport, Thomas H., and Nitin Mittal. "Stop Tinkering with AI." Harvard Business Review, January-February 2023.
[vi] Ontario Institute for Cancer Research. "Using AI to Connect Cancer Patients with Cutting-Edge Clinical Trials." Ontario Institute for Cancer Research, February 27, 2024
[vii] Zhang, B., Zhang, L., Chen, Q. et al. Harnessing artificial intelligence to improve clinical trial design. Commun Med 3, 191 (2023)
[viii] https://deep6.ai/oncology-clinical-trial-recruitment/
[ix] "Finding the Right Patients for the Right Treatment with AI." Publication Name, https://www.avenga.com/magazine/how-ai-advances-patient-recruitment-in-clinical-trials/
[x] Goldberg, J, Amin, N, Zachariah, K. et al. The Introduction of AI Into Decentralized Clinical Trials: Preparing for a Paradigm Shift. JACC Adv. 2024 Aug, 3 (8). https://doi.org/10.1016/j.jacadv.2024.101094
[xi] Hutson, Matthew. "How AI Is Being Used to Accelerate Clinical Trials." Nature, March 13, 2024.
[xii] McKendrick, Joe. "Needed: Humans to Break Artificial Intelligence Out of Its Silo." Forbes, March 26, 2020
Andrew Spanyi is President of Spanyi International. He is a member of the Board of Advisors at the Association of Business Process Professionals and has been an instructor at the BPM Institute. He is also a member of the Cognitive World Think Tank on enterprise AI.