Designers of precision medicine trials need access to more data sources than traditional trials to refine the parameters of their studies and boost enrollment.
Integrating information from a variety of sources — electronic medical records (EMR), lab results, imaging and molecular and claims data — can help trial designers make decisions about inclusion/exclusion criteria, geographic availability of patients, use of technology, statistical analysis methods and trial cessation.
Of 323 pharma executives responding to an NTT DATA and Informa Pharma Intelligence survey, 81 percent cited understanding the patient population as the most important factor in developing a precision medicine program. To achieve understanding, trials need to integrate information from patient records, other trials’ data and eligible investigators, said Mark Philhower, senior director of NTT DATA’s healthcare and life science industry vertical.
“Evaluating solutions that integrate information across multiple data sources” was the second most important factor, with 11 percent of respondents.
Geographic location was the most significant factor in defining inclusion/exclusion criteria for trial recruitment, according to 66 percent of survey respondents. “Too often, study participants are narrowly defined, but geographically diffused,” Philhower said.
The second most important factor to consider in recruiting was medical/disease condition or progression, cited by 24 percent of respondents. Genetic composition has the biggest impact on progression, said 61 percent, followed by disease pathogenesis (19 percent), patient behavioral characteristics (12 percent) and social circumstances (8 percent).
The most valuable data sources for measuring disease progression are radiology data, clinical data (lab data, EMRs, genomic data) and pathology data, respondents said. Information on responsiveness to therapy and the number of lines of therapy used also is important, they said.
EMRs are the key source for identifying clinical trial populations, 58 percent of respondents said. Physician referrals, laboratory diagnostics and healthcare claims scored more modestly.
Retaining patients in precision medicine trials relies most heavily on patient behavior, according to half of respondents. Efforts to influence behavior positively include patient counseling, community and family coordination, and provision of resources, such as lodging, travel and meals.
Questioned on specific technologies they use to improve patient retention, 83 percent of respondents referred to such traditional communication channels as email, text or voicemail messages. Survey responses showed some evidence of emerging digital technologies coming into play, however, with virtual digital assistants, such as Amazon’s Alexa and Google Assistant, scoring 6 percent, and social robots like Jibo, Pepper and Aibo 1 percent.
The survey also shed light on the potential for using machine learning in precision medicine trials and therapeutic programs.
“The new [artificial intelligence (AI)] models have potential to offer preventive, diagnostic and treatment options and alternatives to healthcare practitioners, supplementing their expertise and insights at a resolution and scale unimaginable just a few years ago,” said Philhower. “The challenge will be to translate and apply these discoveries to affect precision diagnosis and precision therapy for improved patient outcomes.”
Among the emerging technology drivers for precision medicine, statistical analytics was ranked as the top priority for 70 percent of survey participants. Statistical packages, such as cluster or regression analysis tools, were favored by 71 percent of respondents. Excel-based modeling tools followed with 17 percent and AI/machine learning at 9 percent.
Forty-one percent of respondents in data management and analytics roles preferred empirical and system-progression models for quantifying disease progression, followed by semi-mechanistic models at 16 percent. Clinical operations specialists, however, favored clinical (33 percent), laboratory (27 percent), pathology (20 percent) and radiographic methods (19 percent).
To read the survey, click here: https://bit.ly/2vzGNiG.