Ambient Clinical Intelligence: What It Means for the EHR Industry

What is ambient clinical intelligence?

Ambient intelligence refers to physical spaces that are sensitive and responsive to the presence of humans, according to an article published by the National Center for Biotechnology Information (NCBI).

The technology hinges on data collected by sensors and processors imbedded into everyday objects and utilizes machine learning algorithms for data analytics.

For example, your Google Assistant and Amazon Alexa—devices that automatically respond to a person's voice—use ambient intelligence.

Ambient intelligence also has several use cases that could potentially improve patient care and mitigate clinician burden in healthcare settings.

Ambient clinical intelligence leverages contact-based wearable devices and contactless sensors embedded in healthcare settings to collect information like audio data and imaging data of physical spaces, according to research published in ScienceDirect. Machine learning algorithms then efficiently interpret the data for clinical application.

What does ambient clinical intelligence mean for the EHR business?

A recent survey found that 36 percent of clinicians spend more than half their day on administrative tasks in the EHR. What’s more, 72 percent of clinicians expect the time they spend on administrative tasks to increase over the next 12 months.

Current strategies to mitigate clinician burden include medical scribes and medical dictation software, according to a review article published in Nature. However, both options face challenges; scribes are costly to train and have high turnover, and medical dictation software is traditionally limited to the post-visit report.

Ambient intelligence could help alleviate clinician burden related to EHR documentation tasks. A study highlighted in the Nature review revealed that one healthcare provider’s time spent on clinical documentation dropped from two hours to 15 minutes with the implementation of ambient intelligence through microphones attached to eyeglasses.

Implementing ambient intelligence could also lead to more accurate transcription than current EHR documentation aids such as medical scribes. Another study highlighted in the Nature review revealed that a deep-learning model demonstrated a word-level transcription accuracy of 80 percent compared to medical scribes that average 76 percent transcription accuracy.

Ambient clinical intelligence could also be used to support clinical care by monitoring patient health status and care trajectory. Hypothetically, data collected by ambient clinical intelligence systems could be integrated into EHR workflows to give a more complete picture of patient health for clinical decision support and care coordination.

For instance, healthcare organizations could leverage contactless, ambient sensors for continuous and nuanced monitoring of ICU patient mobility to improve care outcomes.

ICU-acquired weaknesses are common in critically ill patients and can lead increased mortality and higher hospital costs. The Nature review cited studies showing that early patient mobilization can reduce the relative incidence of ICU-acquired weaknesses by 40 percent.

The standard mobility assessment requires in-person observation, which is limited by cost impracticality, observer bias, and human error.

The review authors noted that healthcare organizations have used wearable devices to monitor patient movement in the ICU, but this technology has limitations. While wearables can detect pre-ambulation maneuvers, such as the transition from sitting to standing, they cannot detect external assistance or interactions with the physical space, while ambient intelligence can.

Challenges to ambient clinical intelligence implementation

While ambient clinical intelligence has the potential to improve healthcare quality, the continuous collection of sensor data in healthcare settings presents several ethical challenges in terms of privacy, data management, bias, and informed consent.

“Navigating these ethical issues is crucial not only for the success of individual uses, but for acceptance of the field as a whole,” the ScienceDirect article noted.

The potential for bias in AI systems stands as an implementation barrier for ambient clinical intelligence.

“Machine learning processes for artificial intelligence generally use a massive set of data input to produce the desired output by finding patterns in the data,” the authors explained. “Using large amounts of data, the artificial intelligence model is trained to identify patterns and create rules that adjust and improve the model's parameters.”

While developers have created various methods in attempts to mitigate bias during model training, machine learning systems can unintentionally produce bias.

“If the dataset does not reflect the relevant qualities of the population to which the algorithm is applied, then there can be bias in the outcomes,” the study authors explained. “The absence of sufficient geographical distribution of patient cohorts used to train algorithms is another potential source of systemic bias.”

The authors also emphasized that the use of ambient intelligence in clinical settings for research purposes presents the need for proper stewardship of data.  

“A key tenet of privacy in research on humans is stewardship of the data,” the study authors emphasized. “Effective stewardship includes ensuring that only members of the research team have access to the study data, that members of the research team are trained in the areas of data privacy and security and have signed privacy agreements with the sponsoring institutions, and that data practices include minimizing access to fully identifiable data as much as possible.”

Additionally, participants in ambient clinical intelligence research projects must give informed consent like other kinds of clinical research.

“In considering whether to participate in the project, people would need to be aware of the potential use of their data, including how their data might be used for the specific research project underway, future research efforts, and potential collaborations with other investigators,” they noted.

With careful attention to these ethical concerns, the use of ambient clinical intelligence could improve clinician burden and advance clinical research.