Opioid overdose detection using smartphones

See allHide authors and affiliations

Science Translational Medicine  09 Jan 2019:
Vol. 11, Issue 474, eaau8914
DOI: 10.1126/scitranslmed.aau8914
  • Fig. 1 Converting a smartphone into an active sonar monitoring system.

    (A) Schematic of device operation. The smartphone’s speaker transmits an inaudible, custom FMCW signal (red arrow), which is reflected by the subject (blue-dashed arrow) and recorded using the smartphone’s microphone. (B) The reflections arrive at time delays ∆ti and ∆te during inhalation and exhalation; the changes translate to unique frequency shifts ∆fi and ∆fe. (C) The frequency shifts can be estimated by taking an FFT over 15 chirps; the breathing signal is found in a frequency bin corresponding to the subject’s distance from the smartphone. Motion in the environment from a different distance would appear at a different frequency bin and hence can be separated.

  • Fig. 2 Breathing rate accuracy across different scenarios.

    The system was evaluated across (A) different smartphone models, (B) orientations of the smartphone, (C) various positions of the smartphone with respect to (w.r.t.) the subject, (D) in the presence of interference from another nearby moving subject, (E) with environmental noise from devices placed 75 cm from the subject, (F) as a function of distance from the smartphone, and (G) recalibration accuracy after the subject changes the orientation of the phone, as well as slouches. (H) The fraction of time the subject’s respiration and other motion was captured by the algorithm in the SIF deployment. Individual data for benchmark testing are reported in table S2.

  • Fig. 3 Measurement of real-world, high-risk opioid use events.

    (A) Overview of experimental sequence. Testing was performed in an SIF (InSite). (B and C) Sensitivity and specificity for respiratory depression (RD) and central apnea events (CAE). (D) Postinjection respiratory rate on a smartphone versus reference standard. (E) Postinjection central apnea event detection on a smartphone versus reference standard. (F to I) Smartphone breathing signals (amplitude over time) of four subjects whom InSite staff physically checked on after injection. (F) Overdose: Subject exhibits multiple central apnea periods, followed by deep breath. (G) Overdose: Subject exhibits respiratory depression with an average respiratory rate of four breaths per minute. (H) Intervention: Subject exhibits central apnea, is aroused by staff, and becomes agitated. (I) Intervention: Subject slouches after injection and staff physically check on patient, whose breathing does not meet respiratory depression or central apnea thresholds.

  • Fig. 4 Distribution of observed central apnea events in the SIF.

    (A) Histogram of central apnea events per participant. (B) Histogram of the duration of the central apnea events identified by the smartphone-based system.

  • Fig. 5 Measurement of simulated overdose events in the OR.

    (A) Illustration of OR setup. The phone is placed within 1 m of the patient on a surgical (Mayo) stand. A respiratory impedance monitor is fitted around the patient’s chest to measure the true respiratory rate and apnea status. Healthy participants wearing all standard OR monitors have general anesthesia induced. (B) Comparison of time to detection of simulated overdose based on algorithm-identified respiratory failure onset (smartphone) versus real-time detection by the reference standard.

  • Table 1 Demographic summary of participants in algorithm evaluation (SIF and OR).

    SIF, supervised injection facility; OR, operating room.

    SIF (n = 94)Age (years)43 ± 11
    Height (cm)178 ± 8.3
    Weight (kg)77 ± 12.4
    Male, n (%)80 (85)
    Female, n (%)14 (15)
    Caucasian, n (%)62 (66)
    Black, n (%)2 (2)
    First Nations, n (%)30 (32)
    Drug injected
    Fentanyl, n (%)17 (19)
    Heroin, n (%)64 (68)
    Morphine, n (%)12 (12)
    Hydromorphone, n (%)1 (1)
    OR (n = 20)
    Age (years)33 ± 10.8
    Height (cm)174 ± 9
    Weight (kg)75 ± 14.6
    Male, n (%)8 (40)
    Female, n (%)12 (60)
    Caucasian, n (%)16 (80)
    African American, n (%)2 (10)
    Asian, n (%)1 (5)
    Pacific Islander, n (%)1 (5)

Supplementary Materials

  • This PDF file includes:

    • Materials and Methods
    • Fig. S1. Connecting overdose victims with EMS.
    • Fig. S2. Postinjection respiratory rate decrement.
    • Table S1. InSite institutional overdose indicators.
    • Table S2. Benchmark testing and individual data for Fig. 2.
    • Reference (57)

    [Download PDF]

Stay Connected to Science Translational Medicine

Navigate This Article