Before it could be changed you'd need to define exactly mathematically what you want. Note that it can not rely on any external information and all nodes must return the same answer to be able to verify.
"the supply of Bitcoins must grow in proportion to the total value of transactions undertaken using the system" is a handwaving description and personally I think it's impossible to determine that in a rational objective way.
Hand waving not necessary.
Institute a txn fee equal to 0.1% of all sends. The total volume of sends per unit time (velocity) would then be proportional to fee collection. Peg difficulty to achieve a coin generation rate equal to a fraction of total txn fee collection. Keep coin generation going forever. In the event of a fall in difficulty relative to a target trend (e.g. difficulty is set by a central planner to grow at a target of 50% year-on-year forever), destroy all coins collected using the txn fee. This causes a net reduction in money supply and serves as a credible commitment to fight price inflation. If difficulty rises faster than a target trend, distribute the txn fees to miners. In this case there is no need to fight price inflation.
This system uses very limited mining incentives compared to bitcoin. How can it keep miners honest? Require all miners to hold coins in escrow until they fulfill a mining contract with the blockchain. Cap mining returns (txn fees + coin generations) at a fixed fraction of coins held in escrow (say 1%). Release the coins from escrow six months after the miners have used up their entire mining allowance (in other words fulfilled their contract). Anyone wanting to monkey around with major hashing power would have to first establish a huge, long-term financial stake in the success of the system.
For initial distribution just grandfather in everyone currently holding bitcoins. Allow them to download a new client and require them to transfer money to new accounts within a set time limit. If they fail to do this before the deadline, then delete their money.
PS: It is true that having strong tools to fight both deflation and inflation simultaneously is difficult. Here I am just focused on the risk of price inflation which I think is the real problem.