The way I see it, the solution has to be multi-tiered. Soft nanites and hard nanites respond to treatments differently, and the safety of the host (victim) is paramount. The book mentions that doing dialysis succeeds in removing most of the soft nanites in a CFD victim. It's not all, sure, but it's a lot. Also, using a massive EMP will disrupt some, if not most, of the hard nanites. So here's my proposed treatment regimen:
Place the host and treatment operation inside a Faraday cage. Perform dialysis on the host, purging many of the soft nanites in the process. Follow the dialysis with a massive EMP surge, just under the threshold of the host. This will disrupt or destroy many of the hard nanites. Perform another dialysis on the host in order to flush the disrupted nanites out of the host. This second dialysis must be done exceedingly fast, which runs the risk of destroying the host's veins and arteries by crushing them. This must be red-lined though, in order to get as many of the nanites out of the host as possible. Be aware throughout that when a host is nearly dead, the CFD seems to take over completely and get the host back on its feet. So there is a fine line between pushing the host to the edge and pushing the host over the edge.
My worry with this treatment is that there will be times when the host cannot tolerate as much treatment as other hosts, and there will be mistakes. Additionally, there is always the fear that some of the nanites might have survived. Therefore, I propose that this treatment take place every day or two over the course of a month.