My argument with your radiotherapy machine is that we're currently using the 'model that on average saves lives', and therefore anything we can replace it with

My argument with your radiotherapy machine is that we're currently using the 'model that on average saves lives', and therefore anything we can replace it with that 'on average saves slightly _more_ lives' is an improvement.