AI is enabling the rise of sophisticated digital therapeutics
Ben Hargreaves explores the capability of AI and machine learning to produce ever more accurate and effective digital therapeutics. With this type of therapy’s potential to save costs and provide a non-pharmaceutical intervention, digital therapeutics are projected to become a crucial part of healthcare systems in the years to come.
Artificial intelligence has become one of the biggest discussion points across all industries in the last year. The life sciences industry is no stranger to the technology, and has been working with the potential of AI and machine learning for a number of years. The focus has been on some of the regular pain points that plague the industry, such as achieving manufacturing efficiencies or to streamline drug development. However, there is one area that has gone beyond looking at easing existing challenges, and into the area of providing new, active solutions, in the form of digital therapeutics.
The digital therapeutics space has grown rapidly since the first US FDA approval for such a product was made in 2017. The ability to support patients with a standalone, non-pharmaceutical means of therapy has significant advantages, whilst also being able to provide additional benefits when used in conjunction with existing treatments. AI and machine learning have been a major part of this wave of new digital therapeutics since they first emerged and, as the technology improves, this is expected to confer even greater benefits.
In practice
The term ‘digital therapeutics’ encompasses a broad umbrella of tools that provide various healthcare services – from virtual reality therapies to cognitive behaviour therapies. The reason that the market for such therapeutics has advanced so rapidly is due to a combination of factors: wide adoption of smartphones, the potential to reach patients at home, the efficacy of the developed solutions, and the advancement of AI and machine learning technology.
On the latter technology, there are many potential applications for AI to aid the creation and application of a digital therapeutic. The technology can be used to track patient data to identify potential health risks, and there is the possibility to combine the therapeutics with wearable technology to improve the level of data being generated. With this data, there is the ability to understand and utilise biomarkers to analyse the efficacy of a treatment. Digital therapeutics can also be used to manage addiction by providing prompts and alerts to help patients with cravings when the AI model has identified an appropriate time to deliver the message.
One company that has an approved digital therapeutic designed to help patients manage nicotine cravings is Click Therapeutics, while the company also has a pipeline of therapeutics targeting schizophrenia, major depressive disorder, and other conditions. The company’s marketed digital therapeutic in smoking cessation uses machine learning to understand the trigger for smoking, or the motivation logged, in order to help the patient achieve their goal. According to the company, the app helped 45% of Clickotine users be smoke-free after eight weeks of a clinical study.
Chief technology officer of Click Therapeutics, Han Chiu, told pharmaphorum of the advantages to the digital therapeutic: “Accessible through a patient’s smartphone, Click’s evidence-based mobile applications are designed to provide safe and effective treatment with fewer side effects than existing pharmacotherapy options. Click employs cutting-edge technical capabilities and rigorous clinical validation, with the goal of developing meaningful clinical interventions [that] are supported by scientific evidence and are enhanced by AI and machine learning for personalisation and continual improvement.”
In terms of how the company’s digital therapeutics are developed, Chiu outlined that they are created through an AI-enabled platform, which allows for the delivery of personalised treatments to patients. In practice, this means that the more patients engage with their treatment, the more the app is able to adapt to their symptoms and responses, while also aiming to increase engagement to achieve the desired outcomes for the patient. “Our programs are architected to securely collect robust, high density, and longitudinal real world patient data sets to support these AI-based approaches,” Chiu added.
From the ground up
AI is playing a key role in the development of many digital therapeutics from the point of conception. In this way, the technology can differentiate itself from standard treatments that are provided based on in-person consultation, but can instead adapt to patient feedback from the outset. The creation of a feedback loop that develops a more effective product is part of the allure of AI technology.
For Click Therapeutics, its website states that its products are built using the Click Neurobehavioral Intervention platform and powered by Clickometrics, the company’s machine learning and data science engine. By developing its therapeutics through a platform, the company is able to use a broad range of feedback to “predict the optimal treatment journey for each patient.”
Chiu explained: “Click Therapeutics has an in-house end-to-end discovery lab dedicated to inventing and validating novel mechanisms of action. During our research and development process, we ensure that we are developing interventions that are adaptive and learn from our AI-enabled platform. Additionally, AI is a core component of our lifecycle management approach, allowing us to continue to work towards the development of version 2.0, 3.0, etc., as we learn from our products.”
Looking to the future
Many of the discussions that are currently taking placing centred on AI concern the benefits, or risks, that might be brought by the technology in the future. In terms of digital therapeutics, the technology will continue to evolve, as highlighted by Chiu, as the R&D processes incorporate feedback from previous iterations.
Chiu described how he sees this playing out in the field of digital therapeutics: “Machine learning algorithms will continue to become more sophisticated and enable high-fidelity insights from patient data. This real-world evidence will enable the development and deployment of more efficacious treatments over time that are robustly personalised to account for the unique situations that patients encounter.”
Chiu extended the advancement of AI into the potential for sensors to also increase in sophistication, and provide more accurate and varied data. This is already the case in some current digital therapeutics and clinical trials, where data is taken from wearables to inform researchers on the efficacy of treatments. However, as the sensors advance further, this could allow for the exact quantification of a drug or treatment’s effect, predict disease states, and identify opportunities to make treatments more effective.
“We envision a future where AI supports the full treatment experience, from start-to-finish, in an ongoing process of continuous product personalisation and improvement,” Chiu concluded.
