Market: Customer Survey: 57 clinicians from academic medical centers and community hospitals, and 44 industry professionals. Primary Customer Need: 70% of neurologists interviewed reported that the most pressing need for a new product was for more reliable automated seizure detection. Over 50% believed that real-time automated seizure detection would be highly beneficial in optimizing treatment using antiepileptic drugs. Market: Customer Survey: 57 clinicians from academic medical centers and community hospitals, and 44 industry professionals. Primary Customer Need: 70% of neurologists interviewed reported that the most pressing need for a new product was for more reliable automated seizure detection. Over 50% believed that real-time automated seizure detection would be highly beneficial in optimizing treatment using antiepileptic drugs. Summary High performance EEG signal event detection increases the efficiency and quality of care. Big data and machine learning together enable unprecedented levels of performance and improve interrater agreement. Automated analysis increases the potential market for EEG technology and addresses many issues faced in critical care. Future applications include real-time seizure detection for ICU applications, cohort retrieval, and medical student training. Acknowledgements This is a collaboration between the University City Science Center and Temple University. Portions of this work were also sponsored by the Defense Advanced Research Projects Agency (DARPA), National Science Foundation and National Institutes of Health. Summary High performance EEG signal event detection increases the efficiency and quality of care. Big data and machine learning together enable unprecedented levels of performance and improve interrater agreement. Automated analysis increases the potential market for EEG technology and addresses many issues faced in critical care. Future applications include real-time seizure detection for ICU applications, cohort retrieval, and medical student training. Acknowledgements This is a collaboration between the University City Science Center and Temple University. Portions of this work were also sponsored by the Defense Advanced Research Projects Agency (DARPA), National Science Foundation and National Institutes of Health. Solution A clinical decision support tool based on proven, advanced, deep learning technology. Identify EEG events in the signal and summarize findings based on the events detected. This market leading product:  enables clinical neurologists employing a volume-based business model to decrease the time spent analyzing an EEG and thereby increase billing;  allows pharmas to assess changes quantitatively in neural activation during clinical trials;  allows neurologists to order and bill for substantially more long-term monitoring tests based on this proven decision support tool;  adds value to the commodity EEG headsets currently entering the market by providing meaningful, real-time signal analysis. Solution A clinical decision support tool based on proven, advanced, deep learning technology. Identify EEG events in the signal and summarize findings based on the events detected. This market leading product:  enables clinical neurologists employing a volume-based business model to decrease the time spent analyzing an EEG and thereby increase billing;  allows pharmas to assess changes quantitatively in neural activation during clinical trials;  allows neurologists to order and bill for substantially more long-term monitoring tests based on this proven decision support tool;  adds value to the commodity EEG headsets currently entering the market by providing meaningful, real-time signal analysis. AUTOEEG: AUTOMATIC INTERPRETATION OF EEGS Abstract The emergence of big data and deep learning is enabling the ability to automatically learn how to interpret EEGs from big data. The TUH EEG Corpus is the largest and most comprehensive publicly-released corpus representing 14 years of clinical data collected at Temple Hospital. It includes over 15,000 patients, 28,000+ sessions, 50,000+ EEGs and deidentified clinical information. We have developed a system, AutoEEG, that recognizes key EEG signal events, generates time aligned markers indicating points of interest in the signal, and then produces a summarization if its findings based on a statistical analysis. Physicians can view the data from any portable computing device and interactively query the data using standard query tools. Clinical consequences include real-time feedback and decision making support. Abstract The emergence of big data and deep learning is enabling the ability to automatically learn how to interpret EEGs from big data. The TUH EEG Corpus is the largest and most comprehensive publicly-released corpus representing 14 years of clinical data collected at Temple Hospital. It includes over 15,000 patients, 28,000+ sessions, 50,000+ EEGs and deidentified clinical information. We have developed a system, AutoEEG, that recognizes key EEG signal events, generates time aligned markers indicating points of interest in the signal, and then produces a summarization if its findings based on a statistical analysis. Physicians can view the data from any portable computing device and interactively query the data using standard query tools. Clinical consequences include real-time feedback and decision making support. Problem Electroencephalography (EEG) is increasingly being used for preventive diagnostic procedures. A board certified EEG specialist currently interprets an EEG. It takes several years of training to learn this art. Interpreting an EEG is time-consuming and there is only moderate interrater agreement. Trend towards long-term monitoring (LTMs). Problem Electroencephalography (EEG) is increasingly being used for preventive diagnostic procedures. A board certified EEG specialist currently interprets an EEG. It takes several years of training to learn this art. Interpreting an EEG is time-consuming and there is only moderate interrater agreement. Trend towards long-term monitoring (LTMs). Performance Performance target: 95% detection accuracy (Det.) at less than 5% false alarms (FAs). Three signal event classes: spike and/or sharp waves (SPSW), (2) periodic lateralized epileptiform discharges (PLED), (3) generalized periodic epileptiform discharges (GPED). Three non-signal cevent classes: (4) artifacts (ARTF), (5) eye movement (EYEM) and (6) background (BCKG). Performance Performance target: 95% detection accuracy (Det.) at less than 5% false alarms (FAs). Three signal event classes: spike and/or sharp waves (SPSW), (2) periodic lateralized epileptiform discharges (PLED), (3) generalized periodic epileptiform discharges (GPED). Three non-signal cevent classes: (4) artifacts (ARTF), (5) eye movement (EYEM) and (6) background (BCKG). System Overview Hybrid hidden Markov models/deep learning: Live-input demonstration: Big Data Resources: Active learning used during training due to the limited amount of transcribed data: System Overview Hybrid hidden Markov models/deep learning: Live-input demonstration: Big Data Resources: Active learning used during training due to the limited amount of transcribed data: B IO S IGNAL A NALYTICS, I NC. Joe Camaratta BioSignal Analytics, Inc. University City Science Center Iyad Obeid and Joseph Picone The Neural Engineering Data Consortium Temple University TAM $1T SAM $1.5B Target $10M Customers: Manufacturers End-Users: Neurologists Competitors: Third-Parties Purchasers: Administrators SystemDet.FAsError Heuristics99%64%74% Random Forest85%6%37% P184%4%37% P1+P282%4%39% P1+P2+P389%4%36%

FieldDescriptionExample 1Version Number0 2Patient IDTUH GenderM 4Date of Birth57 8Firstname_LastnameTUH Startdate01-MAY Study Number/ Tech. IDTUH /TAS X 14Start Date Start Time Number of Bytes in Header Type of SignalEDF+C 19Number of Data Records207 20Dur. of a Data Record (Secs)1 21No. of Signals in a Record24 27Signal[1] PrefilteringHP:1.000 Hz LP:70.0 Hz N: Signal[1] No. Samples/Rec.250 DescriptionExample GenderM (46%), F (54%) Age (Derived from DOB) Min (20), Max (94) Avg (53), Stdev (19) Duration42 hours (17 mins./study) Number of Channels 28 (2%), 33 (15%), 34 (23%) 37 (11%), 42 (29%), 129 (3%) PrefilteringHP:0.000 Hz LP:0.0 Hz N:0.0 Sample Frequency250 Hz (100), 256 Hz (43) Numeric LabelName 1Hyperventilation 2Movement 3Sleeping 4Cough 5Drowsy 6Talking 7Chew 8Seizure 9Swallow 10Spike 11Dizzy 12Twitch MarkerFrequency Eyes Open38% Eyes Closed28% Movement17% Swallow7% Awake4% Drowsy / Sleeping3% Hyperventilation2% Talking1% Market EEG signal files and reports had to be manually paired, de-identified and annotated: Market EEG signal files and reports had to be manually paired, de-identified and annotated: Solution Machine learning algorithms based on hidden Markov models and deep learning are used to learn mappings of EEG events to diagnoses. The system accepts multichannel EEG raw data files as input. Desired output is a transcribed signal and a probability vector with various probable diagnoses. A simple filter bank-based cepstral analysis is used to convert EEG signals to features. The signal is analyzed in 1 sec epochs using 100 msec frames. HMMs are used to map frames to epochs and classify epochs. Solution Machine learning algorithms based on hidden Markov models and deep learning are used to learn mappings of EEG events to diagnoses. The system accepts multichannel EEG raw data files as input. Desired output is a transcribed signal and a probability vector with various probable diagnoses. A simple filter bank-based cepstral analysis is used to convert EEG signals to features. The signal is analyzed in 1 sec epochs using 100 msec frames. HMMs are used to map frames to epochs and classify epochs. Problem Electroencephalography is increasingly being used for preventive diagnostic procedures. A board certified EEG specialist currently interprets an EEG. It takes several year of training to learn this art. Interpreting an EEG is time-consuming and there is only moderate inter-observer agreement. Problem Electroencephalography is increasingly being used for preventive diagnostic procedures. A board certified EEG specialist currently interprets an EEG. It takes several year of training to learn this art. Interpreting an EEG is time-consuming and there is only moderate inter-observer agreement.