Above is a video demonstrating Google duplex, unveiled at Google's I/O convention (May 8-10, 2018). Start at 2:50 if you are short on time. This is a very advanced application of "narrow" AI, and is remarkable because it shows that narrow AI may pass the Turing test, at least under certain conditions.
is a visual introduction from R2D3. In this basic approach, a decision tree is formed for illustration
purposes, using data on homes in New York City and San Francisco. The principles used here are similar
to those applied in more advanced ML applications.
This video is one the best illustrations available currently on the internet, which in a few minutes provides an introduction to the 'black box' that machine learning is often perceived as.
In the past, ML capabilities were available to primarily to large private, government and academic institutions, but are now becoming widely available (although big organizations maintain many advantages). Many organizations are providing standard platforms, 'apps', and training, e.g. Google, Hortonworks and others.
In the table below, a few organizations that engage in AI are listed alphabetically and described. This is a sample, as the universe of such companies is now quite large, and growing rapidly (please note that listing them doesn't imply endorsement of their products). Formation of partnerships and collaborations is quite frequent (between tech and tech, tech and CROs, etc.)
| Apple | Apple is transforming clinical research and healthcare in a variety of ways. The "ResearchKit" allows clinical researchers to engage with Apple's technology. |
| Converseon | Converseon specializes in social listening, which currently is used more in marketing than in clinical research. However, their technology can be adapted for clinical trial use. |
| Datavant | Datavant aggregates and structures healthcare data to generate actionable insights for clinical trials. It's part of the so called "Vants", founded quite recently, with a disruptive and innovative focus, but still unproven record. |
| Deep6 | Deep6 applies natural language processing (a type of AI) on unstructured electronic health records to find patiients who may be eligible for clinical trials. |
| Duke | A highly esteemed academic institution, Duke now has started their own CRO, and frequently forms partnerships. They organized a conference to develop a way to categorize new forms of data; output of which is linked to the left in this row. |
| In the news lately for the Cambridge Analytica scandal, facebook has made efforts to enable their platform for clinical trial use, and has reached out to researchers. | |
| Fitbit | The bulky 'actigraphs' of the past are still used, but Fitbit is doing its part to change that. Its a competitive space, with other players developing wearables of various sorts for applications in clinical research. Due to this competition, Fitbit's research capabilities are increasing swiftly. |
| Like other tech giants, Google has also begun to dabble in the clinical research arena. Adwords is becoming a solid option to consider as part of a clinical trial recruitment strategy, but Google is doing other things too. | |
| Healthmatchapp | Healthmatch is a startup from Australia that has devloped an app designed for patient use, that matches them with clinical trials based on their responses to machine-generated questions. |
| Ibm | Another giant that has made a commitment towards healthcare and clinical research. The Watson AI platform provides a number of features that are applicable to different phases of a clinical trial. |
| Iqvia | By merging, IMS and Quintiles combined expertise sets in datascience and clinical trial operations, and now can do some things not possible for other organizations. |
| Medable | With a surfeit of firms, apps and an evermore complex technology environment, integrators can serve a need for some organizations. This small firm attempts to do that, and is establishing a presence in moble data capture. |
| Mindstronghealth | Mindstrong health, as the name suggests, focuses on neuropsychiatric disorders that has developed digital phenotypes based on data captured from devices. |
| Oracle | No longer the sharp new kid on the block, Oracle can be considered a workhorse in some ways, powering many processes in the background. Its "Oracle Clinical" focuses on applications for clinical researchers. |
| Science37 | While many firms provide mobile solutions, Science37 provides a nearly complete package that can be used to conduct a siteless trial. |
| Threadresearch | Thread offers e-platform, cloud and telehealth, smartphone apps, ePROs, and other services. Its suite of offerings can support many aspects of siteless trials. |
| Transcelerate | TransCelerate fosters collaborations between pharma companies and other organizations, and develops frameworks and systems to address needs in clinical research. It provides templates and tools on its website listed as "initiative assets". |
| Trinetx | Less than 5 years ago, Trinetx started as a venture fund organization that, using proprietary technology that interacted with medical records, specialized in helping sponsors and sites identify patients likely to meet protocol entry criteria. Its service offerings have since branched quite considerably. |
| Twitter allows for very rapid dissemination of information and can be part of a social media campaign for clinical trial recruitment. | |
| Vivosense | Vivosense is a small organization that specializes in algorithms that analyze the new types of data available from wearables and sensors. |
Podcasts
Trends
Below are a few resources that cover this topic, arranged starting with the more general aspects, then moving to specific. It is not intended to be comprehensive but rather as a starting point. This is becoming a specialized field, which is briskly evolving.
These books cover some general topics that go beyond what we did in our workshop. All the subjects addressed by these books are progressing rapidly, so it is likely that the books are now out of date, particularly with regard to details. They may still provide some insights, and some views of longer-term perspectives.