You can apply three different vulnerability approaches for your case, like social vulnerability index (SVI), livelihood vulnerability index (LVI), and climate vulnerability index (CVI), etc to predict the vulnerability of riverbank erosion affected people.

Since you mentioned specifically river bank erosion, Dr. David Rosgen (Colorado) has a BANCS approach and BEHI inventory method to help evaluate bank erosion rates. He also has a paper on cross vanes and J Hooks as structures to reduce near bank stress such as along river bends. It is also important to know the stream type, whether stable or unstable. Unstable is either actively degrading or aggrading, and either one can contribute to bank instability. Aggrading channels may turn into braided streams, and these frequently flood. Some stream bank materials are more stable than others. Rosgen has various approaches to help measure bank erosion, and channel aggrading or degrading by placing channel chains and recording changes after major events. The Rivermorph software he recommends does well at capturing all his channel survey data, cross sections, linear profiles, BEHI data, and helps with calculations. His approach may require substantial sampling and data collection, especially in physiographic areas where his methods have not been used. Rosgen info at www.wildlandhydrology.com

I am not certain whether his approach could satisfy your specific need to address vulnerability, it may take some added materials or structural engineering or other analysis to address vulnerability to structures, etc. Some degree of watershed, climactic, geologic or hydrologic analysis may also be needed to address potential land instability, erosion, sediment sources, etc. within the watershed or basin.

I could not finish, but there are various bioengineering methods or geophysical methods that are sometimes used to help address bank instability. They don’t always work in the long run, and sometimes due to people cutting down vegetation that supports the bank, or flood may occur before measures have had time to fully stabilize. The USDA NRCS Part 650 Engineering Manual addresses some of these approaches.

It is important to identify the floodplain, and collect historical records of flooding and streamflow record in areas of special interest. There is substantial variability associated with flood prediction and climate change, so with urban and other communities along river banks, identifying the 50, 100 and 500 year flood estimate may be needed. The weather forecasting is generally quite good for general use, but transferring that information to specific circumstances or communities may take added measures. And if there are upstream dams, one must know if they will be releasing water for capacity during flood, or have potential for failing. Some streams and rivers have rapid or flashy response, so is there any early warning system to help reduce vulnerability to downstream floodzone hazards. If the area can be affected by severe and rapid flow changes such as rapid snowmelt from rain on snow, slow moving cyclonic events more common to coastal moisture sources, tsunami, thunderstorms, etc., these may be some of the most vulnerable. There may be a substantial amount of information or data collection involved on the area of interest as well as within the watershed. As I have mentioned in some of my earlier responses, it is recommended that a licensed civil engineer or registered professional review plans to address liability issues that can sometimes emerge, depending on how this information would be used.