Accurate Prediction of Antimicrobial Susceptibility for Point‐of‐Care Testing of Urine in Less than 90 Minutes via iPRISM Cassettes

Abstract The extensive and improper use of antibiotics has led to a dramatic increase in the frequency of antibiotic resistance among human pathogens, complicating infectious disease treatments. In this work, a method for rapid antimicrobial susceptibility testing (AST) is presented using microstructured silicon diffraction gratings integrated into prototype devices, which enhance bacteria‐surface interactions and promote bacterial colonization. The silicon microstructures act also as optical sensors for monitoring bacterial growth upon exposure to antibiotics in a real‐time and label‐free manner via intensity‐based phase‐shift reflectometric interference spectroscopic measurements (iPRISM). Rapid AST using clinical isolates of Escherichia coli (E. coli) from urine is established and the assay is applied directly on unprocessed urine samples from urinary tract infection patients. When coupled with a machine learning algorithm trained on clinical samples, the iPRISM AST is able to predict the resistance or susceptibility of a new clinical sample with an Area Under the Receiver Operating Characteristic curve (AUC) of ∼ 0.85 in 1 h, and AUC > 0.9 in 90 min, when compared to state‐of‐the‐art automated AST methods used in the clinic while being an order of magnitude faster.


Supplementary note 1: Dynamic Time Warping Calculation
Formally, given two signals P and Q of length m and n, respectively, the dynamic time warping distance between them DTW(P,Q) is given as follows: [97] DTW(P,Q) Where the minimization is performed over all valid alignment paths .An alignment path of length K is a sequence of K index pairs and is admissible if and only if it satisfies the following conditions.

Boundary: and
Monotonicity: i i i i and j j j Step size: (

) [ ]
Intuitively, if we construct a cost matrix C, where ‖ ‖ , then the optimal alignment path is the "valley" of lowest cost running from the bottom left to the top right of C.
The boundary condition enforces that the starts at the bottom left corner and ends at the top right corner, the monotonicity condition ensures that does not ever turn "backwards" or cross over itself, and the step size ensures that no "jumps" are made (Figure 6).

Figure S3 .
Figure S3.iPRISM relative growth vs. time curves corresponding upon exposure to different concentrations of Gentamicin, calculated as the fraction of bacteria growth compared to the growth without antibiotics for A) E. coli ATCC 25922 and B) S. aureus ATCC 29213.

Figure S4 .
Figure S4.iPRISM AST assay for A) E. coli ATCC 25922 upon exposure to different concentrations of ciprofloxacin and B) Enterococcus faecalis upon exposure to different concentrations of vancomycin.i) Representative real-time iPRISM curves, depicting changes in (-ΔI) over time upon exposure of the Si microstructures to medium (no bacteria), bacteria (no antibiotics) and bacteria with increasing antibiotic concentrations.ii) Corresponding relative growth over time, calculated as the fraction of bacteria growth compared to the growth without antibiotics.iii) Averaged relative growth values at 90 min after exposure to different antibiotic concentrations (n ≥ 6, *** and **** indicate a statistically significant difference compared to bacteria without antibiotics [0 µg mL -1 ] with p < 0.001 and p < 0.0001, respectively).

Figure S5 .
Figure S5.Real-time iPRISM curves for A) susceptible and B) resistant S. aureus clinical isolates from BSI upon exposure to different concentrations of gentamicin.The results are presented as -ΔI values vs. time (left) and the corresponding calculated relative growth values (right) (n=3).

Figure S6 .
Figure S6.iPRISM AST assay demonstration for E. coli upon exposure to ciprofloxacin breakpoint concentration of 0.06 µg mL -1 .A) iPRISM relative growth values at 90 min (RG 90 min ) for 27 susceptible or resistant E. coli clinical isolates from UTIs (n ≥ 2 for each data set) and B) corresponding box plot showing RG 90 min = 0.95 as a threshold to differentiate between resistant and susceptible bacteria.

Figure S7 .
Figure S7.Direct iPRISM AST assay in clinical human urine samples.A) Representative real-time iPRISM growth curve of urine sample spiked with E. coli ATCC 25922, upon exposure to different concentrations of gentamicin.B) Comparison of the iPRISM relative growth values at 90 min (RG 90 min ) values for E. coli ATCC 25922 spiked in growth medium or in urine sample, upon exposure to different concentrations of gentamicin (n ≥ 3 for each data set).

Figure S8 .
Figure S8.Direct iPRISM AST assay in clinical human urine samples for E. coli upon exposure to ciprofloxacin breakpoint concentration of 0.06 µg mL -1 .iPRISM relative growth values at 90 min.Dashed lines indicate the RG 90 min = 0.95 threshold value.

Figure S9 .
Figure S9.iPRISM convergence.As the number of measurements increases, the change in the matrix of pairwise DTW distances X, decreases monotonically.After 100 minutes, the change is negligible.

Table S1 .
Summary of demographic characteristics and pH values of collected clinical urine specimens.Movie of E. coli cells growth on the silicon surface for the first 25 minutes.