Waqaas Al-Siddiq, CEO, Biotricity04.07.17
Although artificial intelligence (AI) is rapidly gaining traction throughout a sundry of vertical industries, its impact on the healthcare space could very well prove the most profound for contemporary society.
Its propensity for automation, machine intelligence, and refined celerity in analyzing disproportionately large and variegated types of data is ideal for transforming the utility of medical device technologies, empowering both patients and care providers alike. AI’s utility for these technologies is predicated on a number of current developments that, once fully matured in the coming years, will enable medical devices to personalize care for individuals from initial diagnosis to ongoing treatment options.
The value AI produces for medical devices greatly hinges on typical hardware concerns of processing speed and storage capacity, both of which are substantially greater today than compared to those of a decade ago. Subsequently, the technological advancements that are most pivotal to AI’s impact on medical device technology include machine size, connectivity, and computing efficiency.
Larger devices that have the requisite storage space and processing capabilities for handling the enormous data quantities are ideal for leveraging the benefits of AI. Smaller devices, like wearables and other mobile options, are reliant upon developments in connectivity—predominantly in the form of the Internet of Things (IoT)—and computer chip speed and size for them to afford those boons. With ubiquitous connectivity, smaller devices will be able to leverage AI through the cloud by offloading data processing.
Size Matters
The adoption rates of medical device technology and AI directly correlate to concerns for processing speed, storage space, and device size predominantly because of AI’s learning prowess. Whether manifest via neural networks, machine learning, deep learning, or some other facet of cognitive computing, AI’s central value proposition is that it improves with both time and the quantity of data it consumes. This capacity for machine intelligence is centered on dynamic algorithms that effectually learn from previous data-based experiences how to better achieve the desired outcome of the information system in which they’re deployed. The more data involved in these IT systems, the more AI’s algorithms can perfect their ability to detect anomalies, identify specific medical conditions, or even issue alerts based on such information. Therefore, device size is a relevant factor today because larger ones are able to encompass the data quantities necessary to constantly learn over time.
Portable devices must offload their data for processing because they lack the size of their larger counterparts. These smaller devices necessitate continuous connectivity for offloading purposes, especially in situations in which they are generating constantly streaming biometric patient data. Connectivity is essential to accessing the broader amounts of data integral to the utility AI delivers in conjunction with medical device technology, such as wearables designed for medication adherence. Current developments in computer chip size and speed will soon make it possible to maintain processing within these smaller devices without the need to offload data. Today, the capability for AI to improve over time with immense data quantities requires a substantial emphasis on device size, connectivity, and processing speed.
Contemporary AI Use
The integration of AI with larger medical devices frequently occurs today in use cases pertaining to initial diagnosis and preliminary analysis of data—both of which exploit AI’s automation capabilities. In the latter instance, AI is instrumental in providing the detection of aberrational data, which technicians can then examine to discern if such data is truly anomalous for a particular patient and his or her medical conditions. Early adopters in healthcare have launched AI to determine whether patients are at risk of death or intubation after hospitalization. Whether utilizing machine learning or some form of deep learning algorithms, AI’s learning capabilities can parse through huge data quantities at high velocities to determine which data might be medically relevant. The expedience at which AI can analyze such copious data amounts is one of the primary windfalls of this approach, as it allows medical personnel to maximize their own specialization and concentrate on more profound problems.
Future AI Use: IoT Connectivity
The more cogent use cases of AI’s utility within the healthcare space arguably pertain to its seamless integration with medical device technology in mobile devices, which offer the autonomy and continuity of data generation which surpasses that of cumbersome stationary devices. Medtronic’s pairing with IBM Watson to predict low blood sugar levels in diabetics is a good example. In particular, the wearables market is saturated with a bevy of gadgets which enable the constant generation of data for salutary purposes. Due to the near real-time constraints requisite to glean insight from such devices in a timely manner, and the immense quantity of data that requires processing, their incorporation of AI technologies necessitates direct, ongoing connectivity—such as that characteristic of IoT.
Virtually all IoT analytics are based on cloud deployments in which endpoint devices regularly offload their remote data to centralized cloud locations. This paradigm is essential to utilizing AI with portable medical devices. It provides a viable means of securely positing such data in an environment where it can be integrated, aggregated, and analyzed with other pertinent data to create a more comprehensive scope of patient data. With this model, AI will run in the cloud in a manner which utilizes data produced from the remote device as well as from other relevant sources. The advantage of this approach lies in the cloud’s benefits which include cheap storage, virtually unparalleled scale, a pay-per use pricing model, and the capacity for elastic computing which provisions resources on demand. The nearly limitless scale of the cloud is ideal for the enormous amounts of Big Data which greatly behoove AI’s learning capabilities. These capabilities position AI as the architecture of choice for IoT and the means of facilitating the incessant connectivity it requires. Moreover, there are a host of cloud deployments with AI options which complement this paradigm as well, making it all but a matter of time before smaller wearable devices leverage AI this way.
Future AI Use: Chip Size
Still, the long-term trajectory for the implementation of AI with wearable medical device technology will inevitably veer towards being self-contained in these endpoint devices. The cloud will be deployed for aggregation and holistic analysis of different data types, but advancements in processing speed and effectiveness will enable mobile devices to utilize AI internally without first offloading data for analysis. The ability to deploy AI algorithms directly on compact, portable units will be facilitated by improved computer chips which are simultaneously becoming both smaller and faster. In this regard, the declining size of computer chips is significant for two reasons. First, the electrical currents from transistors have less distance to travel in smaller chips, which helps to improve the overall pace of processing. Second, the former development coincides with the fact that smaller chips are also able to accommodate greater amounts of transistors, which are an integral part of effecting computations.
Moore’s law states that every few years the number of transistors on computer chips will double. Although Moore’s law will eventually reach the atomic barrier, it’s still increasing the processing speeds of computer chips today. The overall impact is that in the future, these chips (powered by increasing rates of transistors) will become diminutive enough to be placed in mobile devices in quantities that can accommodate the computational ability of larger, stationary medical devices currently capable of directly integrating AI. These advancements will enable such devices to encompass greater quantities of data and parse through them in time to yield some of the benefits of mobile medical device technology. These include issuing notifications in response to real-time analysis of patient data, predicting health events in advance to enable users and medical practitioners to (hopefully) avert them in time, and tailoring data analysis for individual users. As always, the longer these devices are deployed with AI, the better AI becomes at focusing its algorithms to assist patients.
Impacting the Future
The capability of AI to improve over time as it amasses more data and learns and refines its algorithms is well suited for medical device technology. IBM Watson’s lengthy history in cancer research is just one case that typifies this fact. AI technologies have the potential to produce an even greater effect within the healthcare industry by reaching patients directly via mobile devices. Initially, doing so will center on connectivity issues and the need to offload data via the cloud, as most endpoint devices in the IoT do. Developments in computer chip processing speed will result in a situation in which AI functionality is embedded within the devices themselves, profoundly changing the way medical devices work. The impact on the healthcare system will be a change in practice workflows, care delivery, and the integration of devices that make treatment recommendations as opposed to simply providing data for physician consumption.
Waqaas Al-Siddiq is founder and CEO of Biotricity, a biometric remote monitoring solutions company. He is a serial entrepreneur, a former investment advisor and expert in wireless communication technology. He has vast experience through executive roles within start-ups, mid-sized companies, and non-profits. For more information visit Biotricity's website.
Its propensity for automation, machine intelligence, and refined celerity in analyzing disproportionately large and variegated types of data is ideal for transforming the utility of medical device technologies, empowering both patients and care providers alike. AI’s utility for these technologies is predicated on a number of current developments that, once fully matured in the coming years, will enable medical devices to personalize care for individuals from initial diagnosis to ongoing treatment options.
The value AI produces for medical devices greatly hinges on typical hardware concerns of processing speed and storage capacity, both of which are substantially greater today than compared to those of a decade ago. Subsequently, the technological advancements that are most pivotal to AI’s impact on medical device technology include machine size, connectivity, and computing efficiency.
Larger devices that have the requisite storage space and processing capabilities for handling the enormous data quantities are ideal for leveraging the benefits of AI. Smaller devices, like wearables and other mobile options, are reliant upon developments in connectivity—predominantly in the form of the Internet of Things (IoT)—and computer chip speed and size for them to afford those boons. With ubiquitous connectivity, smaller devices will be able to leverage AI through the cloud by offloading data processing.
Size Matters
The adoption rates of medical device technology and AI directly correlate to concerns for processing speed, storage space, and device size predominantly because of AI’s learning prowess. Whether manifest via neural networks, machine learning, deep learning, or some other facet of cognitive computing, AI’s central value proposition is that it improves with both time and the quantity of data it consumes. This capacity for machine intelligence is centered on dynamic algorithms that effectually learn from previous data-based experiences how to better achieve the desired outcome of the information system in which they’re deployed. The more data involved in these IT systems, the more AI’s algorithms can perfect their ability to detect anomalies, identify specific medical conditions, or even issue alerts based on such information. Therefore, device size is a relevant factor today because larger ones are able to encompass the data quantities necessary to constantly learn over time.
Portable devices must offload their data for processing because they lack the size of their larger counterparts. These smaller devices necessitate continuous connectivity for offloading purposes, especially in situations in which they are generating constantly streaming biometric patient data. Connectivity is essential to accessing the broader amounts of data integral to the utility AI delivers in conjunction with medical device technology, such as wearables designed for medication adherence. Current developments in computer chip size and speed will soon make it possible to maintain processing within these smaller devices without the need to offload data. Today, the capability for AI to improve over time with immense data quantities requires a substantial emphasis on device size, connectivity, and processing speed.
Contemporary AI Use
The integration of AI with larger medical devices frequently occurs today in use cases pertaining to initial diagnosis and preliminary analysis of data—both of which exploit AI’s automation capabilities. In the latter instance, AI is instrumental in providing the detection of aberrational data, which technicians can then examine to discern if such data is truly anomalous for a particular patient and his or her medical conditions. Early adopters in healthcare have launched AI to determine whether patients are at risk of death or intubation after hospitalization. Whether utilizing machine learning or some form of deep learning algorithms, AI’s learning capabilities can parse through huge data quantities at high velocities to determine which data might be medically relevant. The expedience at which AI can analyze such copious data amounts is one of the primary windfalls of this approach, as it allows medical personnel to maximize their own specialization and concentrate on more profound problems.
Future AI Use: IoT Connectivity
The more cogent use cases of AI’s utility within the healthcare space arguably pertain to its seamless integration with medical device technology in mobile devices, which offer the autonomy and continuity of data generation which surpasses that of cumbersome stationary devices. Medtronic’s pairing with IBM Watson to predict low blood sugar levels in diabetics is a good example. In particular, the wearables market is saturated with a bevy of gadgets which enable the constant generation of data for salutary purposes. Due to the near real-time constraints requisite to glean insight from such devices in a timely manner, and the immense quantity of data that requires processing, their incorporation of AI technologies necessitates direct, ongoing connectivity—such as that characteristic of IoT.
Virtually all IoT analytics are based on cloud deployments in which endpoint devices regularly offload their remote data to centralized cloud locations. This paradigm is essential to utilizing AI with portable medical devices. It provides a viable means of securely positing such data in an environment where it can be integrated, aggregated, and analyzed with other pertinent data to create a more comprehensive scope of patient data. With this model, AI will run in the cloud in a manner which utilizes data produced from the remote device as well as from other relevant sources. The advantage of this approach lies in the cloud’s benefits which include cheap storage, virtually unparalleled scale, a pay-per use pricing model, and the capacity for elastic computing which provisions resources on demand. The nearly limitless scale of the cloud is ideal for the enormous amounts of Big Data which greatly behoove AI’s learning capabilities. These capabilities position AI as the architecture of choice for IoT and the means of facilitating the incessant connectivity it requires. Moreover, there are a host of cloud deployments with AI options which complement this paradigm as well, making it all but a matter of time before smaller wearable devices leverage AI this way.
Future AI Use: Chip Size
Still, the long-term trajectory for the implementation of AI with wearable medical device technology will inevitably veer towards being self-contained in these endpoint devices. The cloud will be deployed for aggregation and holistic analysis of different data types, but advancements in processing speed and effectiveness will enable mobile devices to utilize AI internally without first offloading data for analysis. The ability to deploy AI algorithms directly on compact, portable units will be facilitated by improved computer chips which are simultaneously becoming both smaller and faster. In this regard, the declining size of computer chips is significant for two reasons. First, the electrical currents from transistors have less distance to travel in smaller chips, which helps to improve the overall pace of processing. Second, the former development coincides with the fact that smaller chips are also able to accommodate greater amounts of transistors, which are an integral part of effecting computations.
Moore’s law states that every few years the number of transistors on computer chips will double. Although Moore’s law will eventually reach the atomic barrier, it’s still increasing the processing speeds of computer chips today. The overall impact is that in the future, these chips (powered by increasing rates of transistors) will become diminutive enough to be placed in mobile devices in quantities that can accommodate the computational ability of larger, stationary medical devices currently capable of directly integrating AI. These advancements will enable such devices to encompass greater quantities of data and parse through them in time to yield some of the benefits of mobile medical device technology. These include issuing notifications in response to real-time analysis of patient data, predicting health events in advance to enable users and medical practitioners to (hopefully) avert them in time, and tailoring data analysis for individual users. As always, the longer these devices are deployed with AI, the better AI becomes at focusing its algorithms to assist patients.
Impacting the Future
The capability of AI to improve over time as it amasses more data and learns and refines its algorithms is well suited for medical device technology. IBM Watson’s lengthy history in cancer research is just one case that typifies this fact. AI technologies have the potential to produce an even greater effect within the healthcare industry by reaching patients directly via mobile devices. Initially, doing so will center on connectivity issues and the need to offload data via the cloud, as most endpoint devices in the IoT do. Developments in computer chip processing speed will result in a situation in which AI functionality is embedded within the devices themselves, profoundly changing the way medical devices work. The impact on the healthcare system will be a change in practice workflows, care delivery, and the integration of devices that make treatment recommendations as opposed to simply providing data for physician consumption.
Waqaas Al-Siddiq is founder and CEO of Biotricity, a biometric remote monitoring solutions company. He is a serial entrepreneur, a former investment advisor and expert in wireless communication technology. He has vast experience through executive roles within start-ups, mid-sized companies, and non-profits. For more information visit Biotricity's website.