TABLE 1.

Drug killing rates for the continuous-time and discrete-time models^a

Drug	Half-life	Stage specificity	Continuous-time model	Discrete-time model
Hypothetical drug 1	Long	No	Vmax = ln(PRR48)/48 + a	V′max = ln(PRR48)/48 + a
Partner drug	Long	Yes	V̂max = ln(PRR48)/48 + a	V̂′max = V̂max 48/q
Hypothetical drug 2	Short	No	Ṽmax = [ln(PRR48) + 48a]/ta	Ṽ′max = [ln(PRR48) + 48a]/ta
Artemisinin derivative PRR48 calibration	Short	Yes	max,48 = [ln(PRR48) + 48a]/ta	′max,48 = max,48 48/q
Artemisinin derivative PRR96 calibration	Short	Yes	max,96 = [ln(PRR48) + 96a]/3ta	Obtained by iteration

^aa is the instantaneous parasite growth rate over the 48-h parasite RBC cycle, PRR₄₈/PRR₉₆ is the reduction in parasite number over 48 or 96 h (i.e., one or two parasite RBC cycles) following drug treatment (the value is different for each drug but identical for both models when used for the same drug), q is the number of 1-h bins during which killing occurs, and t_a is the duration of drug action after each dose. Shown are the equations required to convert the discrete-time model to its continuous-time equivalent for a single patient, i.e., to match the maximal parasite kill rate (V_max in equation 3) in the instantaneous model to its equivalent V′_max value in the discrete-time model (equation 4), the latter being denoted by the prime symbol. The circumflex or tilde above V_max indicates whether adjustment has been made for the effects of stage specificity or short half-life, respectively, to compensate for the lack of drug killing in nonsensitive stages and times when the drug is not present during the 48-h (or 96-h) census period.