We’re interested in the application of artificial intelligence to the advancement of clinical healthcare in diagnostics, predictive analytics, and personalized bioelectronic medicine.
Optimization of stimulus waveforms for electroceutical devices
There has been growing interest in the field of electroceuticals, devices that use electrical stimulation as a therapy, across diverse medical disciplines. Most of these devices currently use a standardized waveform (e.g. the shape of the electrical current), and clinicians empirically tune these waveforms by adjusting a few parameters (e.g. amplitude, frequency). One of the major challenges in developing these electroceutical protocols is being able to find optimized waveforms (beyond rectangular biphasic pulses) in order to maximize health outcomes while minimizing adverse effects. In our lab, we explore different techniques and algorithms to try and find patient-specific, energy-efficient waveforms for electroceuticals.
Multimodal alarm systems in intensive care units
Alarm fatigue is a major problem in healthcare institutions across the nation, especially in the critical care setting. Most patients are being monitored by multiple highly sensitive alarms that alert clinicians when certain thresholds are crossed. However, studies have shown that anywhere from 60-90% of these alarms are false alarms or at least clinically insignificant. Large numbers of these alarms have led to decreased quality of care from clinicians due to poorer response times, disruption of critical work flow processes, and desensitization while also increasing patient delirium rates due to noise pollution. Developing our own data warehousing system to capture ICU time series data directly from the monitors here at Dell Medical, we are looking to integrate the data across multiple different streams in order to reduce alarm fatigue among clinicians.
Intelligent algorithms for clinical diagnosis, predictive analytics and personalized therapeutics
With the adoption of electronic health record systems around the country, we stand at the foot of a mountain of clinical data, much of which is not utilized. With recent advances in machine learning and artificial intelligence, this data can be mined more easily and quickly to produce tools that can support clinicians in more meaningful ways. Clinicians are often the ones who understand the problems the best, while engineers understand the tools and algorithms available out there. Here, we hope to bridge that gap between clinicians and engineers and develop tools and theories that can better address some of the biggest challenges in the field today.